{"id":155,"date":"2026-09-08T00:13:12","date_gmt":"2026-09-07T15:13:12","guid":{"rendered":"https:\/\/localmodelwatch.tsuchitsuchi.com\/2026\/09\/08\/minicpm5-2b-released\/"},"modified":"2026-11-03T16:08:17","modified_gmt":"2026-11-03T07:08:17","slug":"minicpm5-2b-released","status":"publish","type":"post","link":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/08\/minicpm5-2b-released\/","title":{"rendered":"MiniCPM5-2B Text Generation Model: Our Test Answers, 4GB+ VRAM"},"content":{"rendered":"<p><!-- lmw:facts --><\/p>\n<h2>At a Glance<\/h2>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Item<\/th>\n<th>Value<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Repository<\/td>\n<td><a href=\"https:\/\/huggingface.co\/openbmb\/MiniCPM5-2B\">openbmb\/MiniCPM5-2B<\/a><\/td>\n<\/tr>\n<tr>\n<td>Family guide<\/td>\n<td><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/model-openbmb-minicpm5-en\/\">MiniCPM5 guide (1 articles)<\/a><\/td>\n<\/tr>\n<tr>\n<td>Publisher guide<\/td>\n<td><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/publisher-openbmb-en\/\">OpenBMB: models and licenses<\/a><\/td>\n<\/tr>\n<tr>\n<td>Published<\/td>\n<td>2026-09-06<\/td>\n<\/tr>\n<tr>\n<td>License<\/td>\n<td>apache-2.0<\/td>\n<\/tr>\n<tr>\n<td>Formats<\/td>\n<td><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/format-safetensors-en\/\">safetensors<\/a><\/td>\n<\/tr>\n<tr>\n<td>Paper<\/td>\n<td><a href=\"https:\/\/arxiv.org\/abs\/2506.07900\">arXiv:2506.07900<\/a>, <a href=\"https:\/\/arxiv.org\/abs\/2602.09003\">arXiv:2602.09003<\/a><\/td>\n<\/tr>\n<tr>\n<td>Source type<\/td>\n<td>Primary source (the publisher itself)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p><em>Values determined by this site&#8217;s code when the information was collected. Dates are JST.<\/em><\/p>\n<p><!-- \/lmw:facts --><\/p>\n<p><!-- lmw:lab-summary --><\/p>\n<p><strong>What we checked ourselves<\/strong><\/p>\n<ul>\n<li>Ran <code>MiniCPM5-2B-Q4_K_M.gguf<\/code> (Q4_K_M) on our CPU-only server (Neoverse-N1, 3 threads, llama.cpp b11223): generation at 13.9 tokens\/s (a CPU figure, not a GPU speed).<\/li>\n<li>Its answers to our fixed Japanese questions are shown verbatim below.<\/li>\n<li>It needs about 3% fewer tokens than the Qwen3 tokenizer for the same Japanese text, so about 1.03\u00d7 as much Japanese fits in the same context length, and generation is faster per character.<\/li>\n<\/ul>\n<p>Details and conditions are in \u201cOur Own Measurements\u201d below.<\/p>\n<p><!-- \/lmw:lab-summary --><\/p>\n<h2>Overview<\/h2>\n<p>OpenBMB has released &#8220;MiniCPM5-2B&#8221;, a new 2B-class on-device model. Following the &#8220;MiniCPM5-1B&#8221; in the same series, it adopts the standard LlamaForCausalLM architecture and is designed to deliver superior performance in code reasoning, mathematical reasoning, long-context understanding, tool use, and agent tasks. According to the model card, it achieves SOTA compared to open-source models of the same size.<\/p>\n<h2>Specifications<\/h2>\n<ul>\n<li>Number of parameters: 2,516,756,480<\/li>\n<li>Non-embedding parameters: 1,981,982,720<\/li>\n<li>Number of layers: 42<\/li>\n<li>Attention heads (GQA): 16 for Q, 2 for KV<\/li>\n<li>Context length: 131,072<\/li>\n<li>License: apache-2.0<\/li>\n<\/ul>\n<h2>Performance<\/h2>\n<p>From the evaluation result comparison table published in the model card, some comparison targets have been narrowed down (MiniCPM5-2B, LFM2.5-2.6B, Qwen3.5-2B, Gemma-4-E2B-it, Qwen3.5-4B, granite-4.2-3B), and key figures are extracted and shown below.<\/p>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<p class=\"lmw-table-hint\" style=\"margin:0 0 4px;font-size:0.85em;opacity:0.7;\">\u2192 Scroll horizontally to see all columns<\/p>\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th><\/th>\n<th>MiniCPM5-2B<\/th>\n<th>2B-class Models \/ LFM2.5-2.6B<\/th>\n<th>2B-class Models \/ Qwen3.5-2B<\/th>\n<th>2B-class Models \/ Gemma-4-E2B-it<\/th>\n<th>4B-class Models \/ Qwen3.5-4B<\/th>\n<th>4B-class Models \/ granite-4.2-3B<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Average<\/td>\n<td>53.9<\/td>\n<td>33.2<\/td>\n<td>28.0<\/td>\n<td>24.6<\/td>\n<td>51.1<\/td>\n<td>42.7<\/td>\n<\/tr>\n<tr>\n<td>Code Reasoning<\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<\/tr>\n<tr>\n<td>LiveCodeBench v6<\/td>\n<td>69.1<\/td>\n<td>42.1<\/td>\n<td>20.2<\/td>\n<td>42.9<\/td>\n<td>56.4<\/td>\n<td>58.9<\/td>\n<\/tr>\n<tr>\n<td>LCB-Pro 25Q2 (Easy)<\/td>\n<td>68.0<\/td>\n<td>30.9<\/td>\n<td>10.3<\/td>\n<td>27.1<\/td>\n<td>58.3<\/td>\n<td>54.6<\/td>\n<\/tr>\n<tr>\n<td>LCB-Pro 25Q2 (Medium)<\/td>\n<td>17.5<\/td>\n<td>0.0<\/td>\n<td>0.0<\/td>\n<td>0.0<\/td>\n<td>7.0<\/td>\n<td>5.3<\/td>\n<\/tr>\n<tr>\n<td>OJBench<\/td>\n<td>32.5<\/td>\n<td>11.2<\/td>\n<td>2.6<\/td>\n<td>11.6<\/td>\n<td>24.8<\/td>\n<td>21.8<\/td>\n<\/tr>\n<tr>\n<td>SciCode (wbg)<\/td>\n<td>26.3 \u2020<\/td>\n<td>14.2 \u2020<\/td>\n<td>2.8 \u2020<\/td>\n<td>20.9 \u2020<\/td>\n<td>16.1 \u2020<\/td>\n<td>24.9 \u2020<\/td>\n<\/tr>\n<tr>\n<td>Math Reasoning<\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<\/tr>\n<tr>\n<td>AIME 2025<\/td>\n<td>86.5<\/td>\n<td>41.9<\/td>\n<td>29.6<\/td>\n<td>31.7<\/td>\n<td>78.8<\/td>\n<td>79.4<\/td>\n<\/tr>\n<tr>\n<td>AIME 2026<\/td>\n<td>86.5<\/td>\n<td>45.2<\/td>\n<td>29.0<\/td>\n<td>39.8<\/td>\n<td>82.7<\/td>\n<td>83.5<\/td>\n<\/tr>\n<tr>\n<td>HMMT Feb 2026<\/td>\n<td>63.8<\/td>\n<td>33.7<\/td>\n<td>20.5<\/td>\n<td>17.8<\/td>\n<td>64.0<\/td>\n<td>60.8<\/td>\n<\/tr>\n<tr>\n<td>MATH-500<\/td>\n<td>94.6<\/td>\n<td>89.6<\/td>\n<td>85.8<\/td>\n<td>85.4<\/td>\n<td>99.0<\/td>\n<td>97.0<\/td>\n<\/tr>\n<tr>\n<td>Instruction Following<\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<\/tr>\n<tr>\n<td>IFBench<\/td>\n<td>66.3<\/td>\n<td>59.0<\/td>\n<td>46.0<\/td>\n<td>25.7<\/td>\n<td>59.0<\/td>\n<td>73.0<\/td>\n<\/tr>\n<tr>\n<td>IFEval<\/td>\n<td>86.7<\/td>\n<td>93.4<\/td>\n<td>77.5<\/td>\n<td>31.4<\/td>\n<td>90.2<\/td>\n<td>93.7<\/td>\n<\/tr>\n<tr>\n<td>Multi-IF<\/td>\n<td>71.8<\/td>\n<td>76.8<\/td>\n<td>57.1<\/td>\n<td>40.3<\/td>\n<td>73.6<\/td>\n<td>75.9<\/td>\n<\/tr>\n<tr>\n<td>General Knowledge<\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<\/tr>\n<tr>\n<td>MMLU-Pro<\/td>\n<td>70.8<\/td>\n<td>65.2<\/td>\n<td>64.3<\/td>\n<td>56.0<\/td>\n<td>78.0<\/td>\n<td>65.8<\/td>\n<\/tr>\n<tr>\n<td>MMLU-Redux<\/td>\n<td>84.7<\/td>\n<td>80.0<\/td>\n<td>80.0<\/td>\n<td>71.8<\/td>\n<td>88.7<\/td>\n<td>78.9<\/td>\n<\/tr>\n<tr>\n<td>HLE<\/td>\n<td>8.9 \u2020<\/td>\n<td>6.2 \u2020<\/td>\n<td>2.6 \u2020<\/td>\n<td>4.8 \u2020<\/td>\n<td>9.9 \u2020<\/td>\n<td>6.6 \u2020<\/td>\n<\/tr>\n<tr>\n<td>GPQA-Diamond<\/td>\n<td>70.2 \u2020<\/td>\n<td>55.8 \u2020<\/td>\n<td>45.6 \u2020<\/td>\n<td>43.3 \u2020<\/td>\n<td>77.1 \u2020<\/td>\n<td>55.9 \u2020<\/td>\n<\/tr>\n<tr>\n<td>SuperGPQA<\/td>\n<td>40.8<\/td>\n<td>26.2<\/td>\n<td>38.6<\/td>\n<td>30.3<\/td>\n<td>52.8<\/td>\n<td>39.9<\/td>\n<\/tr>\n<tr>\n<td>Long Context<\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<\/tr>\n<tr>\n<td>AA-LCR<\/td>\n<td>59.0 \u2020<\/td>\n<td>5.3 \u2020<\/td>\n<td>28.7 \u2020<\/td>\n<td>17.0 \u2020<\/td>\n<td>61.0 \u2020<\/td>\n<td>24.3 \u2020<\/td>\n<\/tr>\n<tr>\n<td>NoLiMa<\/td>\n<td>68.1<\/td>\n<td>0.7<\/td>\n<td>17.1<\/td>\n<td>3.9<\/td>\n<td>43.5<\/td>\n<td>5.1<\/td>\n<\/tr>\n<tr>\n<td>LongBenchPro<\/td>\n<td>44.8<\/td>\n<td>23.7<\/td>\n<td>8.2<\/td>\n<td>42.2<\/td>\n<td>58.4<\/td>\n<td>34.8<\/td>\n<\/tr>\n<tr>\n<td>LongBench v2<\/td>\n<td>43.7<\/td>\n<td>30.3<\/td>\n<td>24.9<\/td>\n<td>33.2<\/td>\n<td>47.3<\/td>\n<td>36.0<\/td>\n<\/tr>\n<tr>\n<td>Tool Use<\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<\/tr>\n<tr>\n<td>\u03c4\u00b3-Bench Banking<\/td>\n<td>20.8 \u2020<\/td>\n<td>7.2 \u2020<\/td>\n<td>2.1<\/td>\n<td>3.9<\/td>\n<td>6.8 \u2020<\/td>\n<td>5.6 \u2020<\/td>\n<\/tr>\n<tr>\n<td>\u03c4\u00b2-Bench Telecom<\/td>\n<td>97.1<\/td>\n<td>90.4<\/td>\n<td>69.0 \u2020<\/td>\n<td>20.8 \u2020<\/td>\n<td>92.1 \u2020<\/td>\n<td>40.9<\/td>\n<\/tr>\n<tr>\n<td>BFCL v4<\/td>\n<td>66.6<\/td>\n<td>61.1<\/td>\n<td>43.6<\/td>\n<td>36.6<\/td>\n<td>56.8<\/td>\n<td>52.2<\/td>\n<\/tr>\n<tr>\n<td>Coding Agent<\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<\/tr>\n<tr>\n<td>SWE-bench Verified<\/td>\n<td>46.4<\/td>\n<td>6.0<\/td>\n<td>5.0<\/td>\n<td>2.0<\/td>\n<td>33.6<\/td>\n<td>36.8<\/td>\n<\/tr>\n<tr>\n<td>SWE-bench Pro<\/td>\n<td>14.4<\/td>\n<td>0.6<\/td>\n<td>0.8<\/td>\n<td>0.0<\/td>\n<td>28.2<\/td>\n<td>12.3<\/td>\n<\/tr>\n<tr>\n<td>Terminal-Bench v2.1<\/td>\n<td>8.6 \u2020<\/td>\n<td>4.5 \u2020<\/td>\n<td>3.0 \u2020<\/td>\n<td>0.4 \u2020<\/td>\n<td>25.8 \u2020<\/td>\n<td>13.9 \u2020<\/td>\n<\/tr>\n<tr>\n<td>Search Agent<\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<\/tr>\n<tr>\n<td>BrowseComp-ZH<\/td>\n<td>43.5<\/td>\n<td>9.8<\/td>\n<td>18.2<\/td>\n<td>4.7<\/td>\n<td>39.6<\/td>\n<td>21.1<\/td>\n<\/tr>\n<tr>\n<td>BrowseComp Top100<\/td>\n<td>39.7<\/td>\n<td>13.7<\/td>\n<td>19.3<\/td>\n<td>6.0<\/td>\n<td>33.3<\/td>\n<td>19.0<\/td>\n<\/tr>\n<tr>\n<td>GAIA Text-103<\/td>\n<td>88.7<\/td>\n<td>49.5<\/td>\n<td>47.9<\/td>\n<td>30.1<\/td>\n<td>78.6<\/td>\n<td>57.3<\/td>\n<\/tr>\n<tr>\n<td>General Agent<\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<td><\/td>\n<\/tr>\n<tr>\n<td>GDPval-AA v2<\/td>\n<td>19.6 \u2020<\/td>\n<td>4.5<\/td>\n<td>0.0<\/td>\n<td>0.0<\/td>\n<td>11.7<\/td>\n<td>0.0 \u2020<\/td>\n<\/tr>\n<tr>\n<td>Claw-Gym<\/td>\n<td>59.2<\/td>\n<td>19.3<\/td>\n<td>25.5<\/td>\n<td>31.3<\/td>\n<td>51.6<\/td>\n<td>60.0<\/td>\n<\/tr>\n<tr>\n<td>WildClaw<\/td>\n<td>23.9<\/td>\n<td>10.2<\/td>\n<td>9.2<\/td>\n<td>8.9<\/td>\n<td>17.0<\/td>\n<td>20.0<\/td>\n<\/tr>\n<tr>\n<td>QwenClaw<\/td>\n<td>42.9<\/td>\n<td>19.3<\/td>\n<td>18.2<\/td>\n<td>14.5<\/td>\n<td>37.1<\/td>\n<td>36.4<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p>According to measurements by the publisher, MiniCPM5-2B recorded an average score of 53.9, showing performance that exceeds 2B-class models of the same scale and some 4B-class models. In particular, it reaches high figures in coding and mathematics-related metrics such as LiveCodeBench v6 and AIME 2025. On the other hand, results show it falling behind some other models in specific instruction-following items such as IFEval. It demonstrates superiority in long-context, tool use, and agent-related tasks (such as GAIA Text-103).<\/p>\n<h2>Strengths and Use Cases<\/h2>\n<p>It is designed for local assistants, coding agents, tool-using workflows, and reasoning scenarios. It maintains a compact deployment footprint while featuring native long-context support (131,072 tokens).<\/p>\n<p><!-- lmw:hardware --><\/p>\n<h2>Hardware Requirements<\/h2>\n<p><strong>Estimated requirements (calculated by Local Model Watch)<\/strong> \u2014 2.5B parameters<\/p>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Your VRAM<\/th>\n<th>Quantization<\/th>\n<th>File size<\/th>\n<th>Est. memory needed<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>4GB (laptop iGPU \/ phone class)<\/td>\n<td>Q8_0<\/td>\n<td>2.5GB<\/td>\n<td>3.0GB<\/td>\n<\/tr>\n<tr>\n<td>8GB (RTX 4060 \/ 3060 Ti, etc.)<\/td>\n<td>F16<\/td>\n<td>4.7GB<\/td>\n<td>5.6GB<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p><strong>Inference engine support<\/strong> (architecture name matched against each project&#8217;s own model registry in its source code, checked 2026-11-03): llama.cpp: registered, vLLM: registered, MLX (mlx-lm): registered.<\/p>\n<p><em>Memory estimates add a 20% runtime overhead (KV cache, etc.) to the actual size of the distributed files. Actual usage varies with context length, batch size and inference engine. These figures are computed by this site from file sizes, not published by the model&#8217;s authors. File sizes are measured from the converted build <a href=\"https:\/\/huggingface.co\/openbmb\/MiniCPM5-2B-GGUF\">openbmb\/MiniCPM5-2B-GGUF<\/a>. Compare with other models in our <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/vram-guide-en\/\">VRAM quick reference<\/a>. What the quantization names mean: <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/glossary-quantization-en\/\">glossary<\/a>.<\/em><\/p>\n<p><!-- \/lmw:hardware --><\/p>\n<p><!-- lmw:runnability --><\/p>\n<h2>Can You Run It Locally?<\/h2>\n<p><strong>Runs in Ollama, LM Studio and llama.cpp via a converted build.<\/strong><\/p>\n<p>The publisher ships safetensors, but <a href=\"https:\/\/huggingface.co\/openbmb\/MiniCPM5-2B-GGUF\">openbmb\/MiniCPM5-2B-GGUF<\/a> provides a GGUF build you can use.<\/p>\n<p><strong>License \u2014 <code>apache-2.0<\/code> (Commercial use allowed):<\/strong> Permits commercial use, modification and redistribution. Redistribution requires including the license and stating changes; includes a patent grant.<\/p>\n<p><strong>Compression:<\/strong> the Q8_0 build measures 8.53 bits per weight \u2014 about 53% the size of the original 16-bit weights, calculated by this site from the actual file sizes.<\/p>\n<p><em>Compiled by this site&#8217;s code from the published formats, converted builds we have found, and each engine&#8217;s own model registry. &#8220;Not found&#8221; means we have not seen such a build, not that none exists. License summaries are not legal advice \u2014 check the publisher&#8217;s original terms before relying on them.<\/em><\/p>\n<p><!-- \/lmw:runnability --><\/p>\n<p><!-- lmw:lab --><\/p>\n<h2>Our Own Measurements<\/h2>\n<p>Values we measured ourselves on our server (no GPU) by actually reading and running this model&#8217;s files \u2014 not figures copied from the model card. How we measure, and the results for every model: <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/observations-en\/\">Our Measurements<\/a>.<\/p>\n<h3>What We Found<\/h3>\n<p>MiniCPM5-2B ran comfortably on our CPU-only server (a Neoverse-N1, 3 threads): the 1.45GB Q4_K_M file generated 13.92 tokens per second, about 20.9 Japanese characters per second, with a peak memory of 2.9GB. That is lighter and faster than Qwen3-4B (6.81 tokens per second) and Qwen3-8B (4.57 tokens per second) measured under the same conditions. Its answers to our five Japanese questions, however, were clearly mixed.<\/p>\n<p>It gave the correct 280 yen on the word problem. Asked to summarize in three sentences, it used four. On the question about prefectures that do not border the sea, it did include all 8 correct ones, but also listed 37 that do border the sea, effectively listing most of Japan&#8217;s prefectures.<\/p>\n<p>In the quantization comparison, agreement with the Q8_0 baseline on the most likely next token fell from 90.8% at Q6_K to 89.1% at Q5_K_M, 81.5% at Q4_K_M and 37.4% at Q2_K. No level reached 95%, and only Q6_K cleared 90%. At Q2_K the summary, the word problem and the haiku ran into the 1,024-token limit, and the business email repeated the same line 11 times.<\/p>\n<p>Note that in the other distribution we used for the quantization comparison, neither the Q8_0 baseline nor Q4_K_M answered 280 yen on the word problem, unlike the publisher&#8217;s own Q4_K_M we used for the CPU run. Even at temperature 0 under the same conditions, a different distribution or quantization can change the answer.<\/p>\n<h3>Japanese Token Efficiency<\/h3>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Tokenizer<\/th>\n<th>Tokens per 1,000 Japanese characters<\/th>\n<th>Ratio to the same text in English<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><strong>This model<\/strong><\/td>\n<td><strong>666<\/strong><\/td>\n<td><strong>1.21\u00d7<\/strong><\/td>\n<\/tr>\n<tr>\n<td>Qwen3<\/td>\n<td>688<\/td>\n<td>1.26\u00d7<\/td>\n<\/tr>\n<tr>\n<td>Llama 3.2<\/td>\n<td>744<\/td>\n<td>1.36\u00d7<\/td>\n<\/tr>\n<tr>\n<td>Gemma 3<\/td>\n<td>564<\/td>\n<td>1.03\u00d7<\/td>\n<\/tr>\n<tr>\n<td>gpt-oss<\/td>\n<td>795<\/td>\n<td>1.45\u00d7<\/td>\n<\/tr>\n<tr>\n<td>LLM-jp-3<\/td>\n<td>497<\/td>\n<td>0.85\u00d7<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p>It needs about 3% fewer tokens than the Qwen3 tokenizer for the same Japanese text, so about 1.03\u00d7 as much Japanese fits in the same context length, and generation is faster per character.<\/p>\n<p>Counted with the <code>tokenizer.json<\/code> of <a href=\"https:\/\/huggingface.co\/openbmb\/MiniCPM5-2B\">openbmb\/MiniCPM5-2B<\/a> on a fixed text we wrote ourselves (876 Japanese characters across news, conversation, technical docs, a formal email, travel writing and a recipe) and its English translation. Fewer tokens mean more Japanese fits in the context window.<\/p>\n<h3>Inside the GGUF File<\/h3>\n<p>File: <a href=\"https:\/\/huggingface.co\/openbmb\/MiniCPM5-2B-GGUF\/blob\/main\/MiniCPM5-2B-Q4_K_M.gguf\">MiniCPM5-2B-Q4_K_M.gguf<\/a> (1.45GB, <code>Q4_K_M<\/code>). We read only the header (metadata) of the file, not the weights.<\/p>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Item<\/th>\n<th>Value<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Architecture (as named in the GGUF)<\/td>\n<td><code>llama<\/code><\/td>\n<\/tr>\n<tr>\n<td>Maximum trained context length<\/td>\n<td>131,072 tokens<\/td>\n<\/tr>\n<tr>\n<td>Layers<\/td>\n<td>42<\/td>\n<\/tr>\n<tr>\n<td>Vocabulary size<\/td>\n<td>130,560<\/td>\n<\/tr>\n<tr>\n<td>Chat template<\/td>\n<td>Included (mentions tool calls, has a thinking switch)<\/td>\n<\/tr>\n<tr>\n<td>imatrix<\/td>\n<td>Not recorded in the file<\/td>\n<\/tr>\n<tr>\n<td>Weight types (share of parameters)<\/td>\n<td>Q4_K 78.4% \/ Q6_K 21.6% \/ F32 0.0%<\/td>\n<\/tr>\n<tr>\n<td>Average bits per weight<\/td>\n<td>4.95 bits<\/td>\n<\/tr>\n<tr>\n<td>Embedding \/ output layer type<\/td>\n<td>Q4_K \/ Q6_K<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p>The quant name in the file name describes the file as a whole; in practice layers mix several types. The average bits per weight is the measured file data divided by the number of weights.<\/p>\n<h3>Quality Loss by Quantization (Measured on Japanese Text)<\/h3>\n<p>We downloaded each quantization of <a href=\"https:\/\/huggingface.co\/bartowski\/MiniCPM5-2B-GGUF\">bartowski\/MiniCPM5-2B-GGUF<\/a> one at a time and compared it with Q8_0 (2.50GB) on a fixed Japanese text (the opening of Natsume Soseki&#8217;s <em>Botchan<\/em>, public domain) using llama.cpp&#8217;s KL-divergence mode.<\/p>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<p class=\"lmw-table-hint\" style=\"margin:0 0 4px;font-size:0.85em;opacity:0.7;\">\u2192 Scroll horizontally to see all columns<\/p>\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Quant<\/th>\n<th>File size<\/th>\n<th>Same top token as the baseline<\/th>\n<th>Mean KL divergence<\/th>\n<th>Perplexity vs baseline<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td><code>Q2_K<\/code><\/td>\n<td>0.94GB<\/td>\n<td>37.4%<\/td>\n<td>1.1494<\/td>\n<td>3.274\u00d7<\/td>\n<\/tr>\n<tr>\n<td><code>Q3_K_M<\/code><\/td>\n<td>1.15GB<\/td>\n<td>63.6%<\/td>\n<td>0.3233<\/td>\n<td>1.412\u00d7<\/td>\n<\/tr>\n<tr>\n<td><code>IQ4_XS<\/code><\/td>\n<td>1.36GB<\/td>\n<td>76.9%<\/td>\n<td>0.1129<\/td>\n<td>1.124\u00d7<\/td>\n<\/tr>\n<tr>\n<td><code>Q4_K_M<\/code><\/td>\n<td>1.50GB<\/td>\n<td>81.5%<\/td>\n<td>0.0703<\/td>\n<td>1.072\u00d7<\/td>\n<\/tr>\n<tr>\n<td><code>Q5_K_M<\/code><\/td>\n<td>1.79GB<\/td>\n<td>89.1%<\/td>\n<td>0.0224<\/td>\n<td>1.024\u00d7<\/td>\n<\/tr>\n<tr>\n<td><code>Q6_K<\/code><\/td>\n<td>1.96GB<\/td>\n<td>90.8%<\/td>\n<td>0.0132<\/td>\n<td>1.010\u00d7<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p>No quantization reached 95% top-token agreement.<\/p>\n<p>KL divergence closer to 0 means the next-token probabilities stay closer to the baseline. Conditions: Neoverse-N1, llama.cpp b11223, context 512 tokens, measured 2026-09-29 (JST). <em>Botchan<\/em> is a famous text the model may have seen in training, so read the numbers as a comparison between quantizations of the same model, not as an absolute score.<\/p>\n<h3>Answers by Quantization Level (Verbatim)<\/h3>\n<p>While measuring the quality loss above, we also asked the same five Japanese questions to <code>Q8_0<\/code>, <code>Q4_K_M<\/code>, <code>Q2_K<\/code> of the same model (temperature 0, up to 1024 tokens) and publish the answers unedited, largest file first. We do not grade them.<\/p>\n<p><strong>Summarization<\/strong><\/p>\n<blockquote>\n<p>\u6b21\u306e\u8a18\u4e8b\u30923\u6587\u3067\u8981\u7d04\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n<p>\u307f\u3069\u308a\u5e02\u306f\u4eca\u5e74\u306e\u6625\u3001\u5e02\u5185\u306b\u4e09\u3064\u3042\u308b\u516c\u7acb\u56f3\u66f8\u9928\u306e\u3046\u3061\u6700\u3082\u53e4\u3044\u4e2d\u592e\u56f3\u66f8\u9928\u306e\u6539\u4fee\u3092\u7d42\u3048\u3001\u5341\u6708\u4e00\u65e5\u306b\u518d\u958b\u9928\u3057\u305f\u3002\u6539\u4fee\u306f\u7d04\u4e00\u5e74\u534a\u306b\u308f\u305f\u308a\u3001\u7dcf\u4e8b\u696d\u8cbb\u306f\u7d04\u5341\u4e8c\u5104\u5186\u3002\u3053\u306e\u3046\u3061\u7d04\u56db\u5104\u5186\u306f\u56fd\u306e\u88dc\u52a9\u91d1\u3067\u307e\u304b\u306a\u308f\u308c\u305f\u3002<\/p>\n<p>\u6700\u5927\u306e\u5909\u5316\u306f\u3001\u95b2\u89a7\u5e2d\u306e\u914d\u7f6e\u3067\u3042\u308b\u3002\u4ee5\u524d\u306f\u9928\u5185\u306e\u5ea7\u5e2d\u306e\u5927\u534a\u304c\u5b66\u7fd2\u7528\u306e\u673a\u3067\u3001\u8a66\u9a13\u671f\u9593\u306b\u306a\u308b\u3068\u4e2d\u9ad8\u751f\u3067\u57cb\u307e\u308a\u3001\u9ad8\u9f62\u306e\u5229\u7528\u8005\u3084\u5c0f\u3055\u306a\u5b50\u3069\u3082\u9023\u308c\u304c\u5ea7\u308c\u306a\u3044\u3068\u3044\u3046\u82e6\u60c5\u304c\u6bce\u5e74\u5bc4\u305b\u3089\u308c\u3066\u3044\u305f\u3002\u6539\u4fee\u5f8c\u306f\u5ea7\u5e2d\u6570\u3092\u767e\u4e8c\u5341\u5e2d\u304b\u3089\u4e8c\u767e\u5e2d\u306b\u5897\u3084\u3057\u305f\u3046\u3048\u3067\u3001\u5b66\u7fd2\u7528\u306e\u673a\u3092\u4e09\u968e\u306b\u307e\u3068\u3081\u3001\u4e00\u968e\u306f\u4f1a\u8a71\u304c\u3067\u304d\u308b\u300c\u306b\u304e\u308f\u3044\u30d5\u30ed\u30a2\u300d\u3001\u4e8c\u968e\u306f\u9759\u304b\u306b\u672c\u3092\u8aad\u3080\u300c\u95b2\u89a7\u30d5\u30ed\u30a2\u300d\u306b\u5206\u3051\u305f\u3002\u4e00\u968e\u306b\u306f\u4e88\u7d04\u5236\u306e\u30b0\u30eb\u30fc\u30d7\u5b66\u7fd2\u5ba4\u304c\u56db\u5ba4\u3068\u3001\u98f2\u307f\u7269\u3092\u6301\u3061\u8fbc\u3081\u308b\u30ab\u30d5\u30a7\u30b9\u30da\u30fc\u30b9\u3082\u8a2d\u3051\u3089\u308c\u305f\u3002<\/p>\n<p>\u8535\u66f8\u306e\u9762\u3067\u3082\u898b\u76f4\u3057\u304c\u9032\u3093\u3060\u3002\u8cb8\u51fa\u306e\u5c11\u306a\u3044\u53e4\u3044\u8cc7\u6599\u7d04\u4e09\u4e07\u518a\u3092\u5e02\u5185\u306e\u4ed6\u306e\u9928\u3084\u4fdd\u5b58\u66f8\u5eab\u306b\u79fb\u3057\u3001\u4ee3\u308f\u308a\u306b\u5b50\u3069\u3082\u5411\u3051\u306e\u7d75\u672c\u3068\u3001\u4ecb\u8b77\u30fb\u5b50\u80b2\u3066\u30fb\u4ed5\u4e8b\u63a2\u3057\u3068\u3044\u3063\u305f\u66ae\u3089\u3057\u306b\u95a2\u308f\u308b\u5b9f\u7528\u66f8\u3092\u7d04\u4e00\u4e07\u4e94\u5343\u518a\u5897\u3084\u3057\u305f\u3002\u96fb\u5b50\u66f8\u7c4d\u306e\u8cb8\u51fa\u30b5\u30fc\u30d3\u30b9\u3082\u59cb\u307e\u308a\u3001\u5229\u7528\u767b\u9332\u3092\u3059\u308c\u3070\u81ea\u5b85\u304b\u3089\u4e00\u5ea6\u306b\u4e09\u518a\u307e\u3067\u3001\u4e8c\u9031\u9593\u501f\u308a\u3089\u308c\u308b\u3002<\/p>\n<p>\u5efa\u7269\u305d\u306e\u3082\u306e\u306b\u3082\u624b\u304c\u5165\u3063\u305f\u3002\u7bc9\u56db\u5341\u5e74\u3092\u8d85\u3048\u3066\u3044\u305f\u7a7a\u8abf\u8a2d\u5099\u3092\u5165\u308c\u66ff\u3048\u3001\u5c4b\u4e0a\u306b\u306f\u592a\u967d\u5149\u30d1\u30cd\u30eb\u3092\u8a2d\u7f6e\u3057\u3066\u3001\u9928\u5185\u3067\u4f7f\u3046\u96fb\u529b\u306e\u7d04\u4e8c\u5272\u3092\u307e\u304b\u306a\u3048\u308b\u3088\u3046\u306b\u3057\u305f\u3002\u8eca\u3044\u3059\u306e\u5229\u7528\u8005\u304b\u3089\u8981\u671b\u306e\u591a\u304b\u3063\u305f\u30a8\u30ec\u30d9\u30fc\u30bf\u30fc\u306f\u4e00\u57fa\u304b\u3089\u4e8c\u57fa\u306b\u5897\u3048\u3001\u66f8\u67b6\u306e\u9593\u306e\u901a\u8def\u3082\u4ee5\u524d\u3088\u308a\u4e09\u5341\u30bb\u30f3\u30c1\u5e83\u3052\u3089\u308c\u305f\u3002\u5165\u53e3\u306b\u306f\u6bb5\u5dee\u306e\u306a\u3044\u30b9\u30ed\u30fc\u30d7\u304c\u65b0\u305f\u306b\u8a2d\u3051\u3089\u308c\u3001\u6388\u4e73\u5ba4\u3068\u591a\u76ee\u7684\u30c8\u30a4\u30ec\u3082\u5404\u968e\u306b\u4e00\u3064\u305a\u3064\u7f6e\u304b\u308c\u305f\u3002<\/p>\n<p>\u4e00\u65b9\u3067\u8ab2\u984c\u3082\u6b8b\u308b\u3002\u99d0\u8eca\u5834\u306f\u6539\u4fee\u524d\u3068\u540c\u3058\u56db\u5341\u53f0\u306e\u307e\u307e\u3067\u3001\u9031\u672b\u306b\u306f\u8fd1\u304f\u306e\u9053\u8def\u306b\u9806\u756a\u5f85\u3061\u306e\u8eca\u304c\u4e26\u3093\u3060\u3002\u5e02\u306f\u96a3\u63a5\u3059\u308b\u5e02\u6709\u5730\u3092\u81e8\u6642\u99d0\u8eca\u5834\u3068\u3057\u3066\u958b\u653e\u3057\u305f\u304c\u3001\u6052\u4e45\u7684\u306a\u5bfe\u7b56\u306f\u6c7a\u307e\u3063\u3066\u3044\u306a\u3044\u3002\u307e\u305f\u3001\u958b\u9928\u6642\u9593\u3092\u5e73\u65e5\u306e\u591c\u4e5d\u6642\u307e\u3067\u5ef6\u3070\u3057\u305f\u3053\u3068\u3067\u8077\u54e1\u306e\u8ca0\u62c5\u304c\u5897\u3048\u3066\u304a\u308a\u3001\u8077\u54e1\u56e3\u4f53\u306f\u4eba\u54e1\u306e\u8ffd\u52a0\u3092\u6c42\u3081\u3066\u3044\u308b\u3002<\/p>\n<p>\u5229\u7528\u8005\u306e\u53d7\u3051\u6b62\u3081\u306f\u304a\u304a\u3080\u306d\u597d\u610f\u7684\u3060\u3002\u9031\u306b\u4e8c\u5ea6\u3001\u5b6b\u3092\u9023\u308c\u3066\u6765\u308b\u3068\u3044\u3046\u4e03\u5341\u4ee3\u306e\u5973\u6027\u306f\u300c\u524d\u306f\u5b50\u3069\u3082\u304c\u58f0\u3092\u51fa\u3059\u3068\u5468\u308a\u306e\u76ee\u304c\u6c17\u306b\u306a\u3063\u305f\u304c\u3001\u4e00\u968e\u306a\u3089\u6c17\u517c\u306d\u306a\u304f\u7d75\u672c\u3092\u8aad\u3093\u3067\u3042\u3052\u3089\u308c\u308b\u300d\u3068\u8a71\u3059\u3002\u4e00\u65b9\u3001\u53d7\u9a13\u3092\u63a7\u3048\u305f\u9ad8\u6821\u751f\u306e\u4e2d\u306b\u306f\u300c\u5b66\u7fd2\u5e2d\u304c\u4e09\u968e\u3060\u3051\u306b\u306a\u308a\u3001\u5915\u65b9\u306f\u3059\u3050\u306b\u6e80\u5e2d\u306b\u306a\u308b\u300d\u3068\u4e0d\u6e80\u3092\u6f0f\u3089\u3059\u4eba\u3082\u3044\u308b\u3002\u56f3\u66f8\u9928\u306f\u3001\u5b66\u7fd2\u5e2d\u306e\u7a7a\u304d\u72b6\u6cc1\u3092\u30db\u30fc\u30e0\u30da\u30fc\u30b8\u3067\u78ba\u8a8d\u3067\u304d\u308b\u4ed5\u7d44\u307f\u3092\u5e74\u5185\u306b\u5c0e\u5165\u3059\u308b\u4e88\u5b9a\u3060\u3002<\/p>\n<p>\u518d\u958b\u9928\u306b\u5408\u308f\u305b\u3066\u3001\u50ac\u3057\u3082\u5897\u3048\u305f\u3002\u6bce\u9031\u571f\u66dc\u306e\u5348\u524d\u306b\u306f\u8aad\u307f\u805e\u304b\u305b\u4f1a\u3001\u6708\u306b\u4e00\u5ea6\u306f\u5730\u5143\u306e\u8fb2\u5bb6\u3084\u5546\u5e97\u4e3b\u3092\u8b1b\u5e2b\u306b\u62db\u304f\u300c\u307e\u3061\u306e\u5148\u751f\u300d\u8b1b\u5ea7\u304c\u958b\u304b\u308c\u308b\u3002\u5341\u4e00\u6708\u306b\u306f\u3001\u5e02\u5185\u306e\u4e2d\u5b66\u751f\u304c\u9078\u3093\u3060\u672c\u3092\u7d39\u4ecb\u3059\u308b\u5c55\u793a\u3082\u4e88\u5b9a\u3055\u308c\u3066\u3044\u308b\u3002<\/p>\n<p>\u9928\u9577\u306e\u5c71\u672c\u548c\u5b50\u3055\u3093\u306f\u300c\u672c\u3092\u501f\u308a\u308b\u5834\u6240\u304b\u3089\u3001\u4eba\u304c\u96c6\u307e\u3063\u3066\u904e\u3054\u3059\u5834\u6240\u3078\u5909\u3048\u305f\u304b\u3063\u305f\u3002\u518d\u958b\u9928\u304b\u3089\u306e\u4e00\u304b\u6708\u3067\u6765\u9928\u8005\u306f\u4ee5\u524d\u306e\u4e00\u30fb\u516b\u500d\u306b\u306a\u3063\u305f\u300d\u3068\u8a71\u3059\u3002\u5e02\u306f\u6765\u5e74\u5ea6\u3001\u6b8b\u308b\u4e8c\u3064\u306e\u56f3\u66f8\u9928\u306b\u3064\u3044\u3066\u3082\u5229\u7528\u8005\u3078\u306e\u805e\u304d\u53d6\u308a\u3092\u884c\u3044\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u3092\u691c\u8a0e\u3059\u308b\u65b9\u91dd\u3060\u3002<\/p>\n<\/blockquote>\n<p><em>(Summarize the following article in three sentences. (The article, about 1,200 Japanese characters, describes the renovation and reopening of a city&#8217;s central library: the budget, the new floor layout and seating, changes to the collection and e-book lending, building and accessibility work, remaining issues such as parking and staff workload, users&#8217; reactions, new events, and the city&#8217;s next steps.))<\/em><\/p>\n<p><code>Q8_0<\/code> (baseline) 2.50GB<\/p>\n<pre><code class=\"language-text\">\u307f\u3069\u308a\u5e02\u306e\u4e2d\u592e\u56f3\u66f8\u9928\u306f\u6539\u4fee\u3092\u7d42\u3048\u5341\u6708\u4e00\u65e5\u306b\u518d\u958b\u9928\u3057\u3001\u5ea7\u5e2d\u6570\u3084\u8535\u66f8\u3001\u8a2d\u5099\u306a\u3069\u3092\u5145\u5b9f\u3055\u305b\u307e\u3057\u305f\u3002\u6539\u4fee\u306b\u306f\u7d04\u5341\u4e8c\u5104\u5186\u306e\u8cbb\u7528\u304c\u304b\u304b\u308a\u3001\u56fd\u306e\u88dc\u52a9\u91d1\u3082\u7d04\u56db\u5104\u5186\u3092\u6d3b\u7528\u3057\u307e\u3057\u305f\u304c\u3001\u99d0\u8eca\u5834\u3084\u958b\u9928\u6642\u9593\u306e\u8ab2\u984c\u3082\u6b8b\u3063\u3066\u3044\u307e\u3059\u3002\u9928\u9577\u306f\u6765\u9928\u8005\u6570\u304c\u5927\u5e45\u306b\u5897\u3048\u305f\u3068\u8ff0\u3079\u3001\u6b8b\u308a\u306e\u56f3\u66f8\u9928\u3082\u5229\u7528\u8005\u306b\u805e\u304d\u53d6\u308a\u3092\u91cd\u8996\u3059\u308b\u65b9\u91dd\u3067\u3059\u3002\n<\/code><\/pre>\n<p><code>Q4_K_M<\/code> 1.50GB<\/p>\n<pre><code class=\"language-text\">\u307f\u3069\u308a\u5e02\u306e\u4e2d\u592e\u56f3\u66f8\u9928\u306f\u6539\u4fee\u3092\u7d42\u3048\u5341\u6708\u4e00\u65e5\u306b\u518d\u958b\u9928\u3057\u3001\u7dcf\u4e8b\u696d\u8cbb\u7d04\u5341\u4e8c\u5104\u5186\u306e\u3046\u3061\u7d04\u56db\u5104\u5186\u304c\u56fd\u306e\u88dc\u52a9\u91d1\u3068\u306a\u3063\u305f\u3002\u6539\u4fee\u3067\u306f\u5ea7\u5e2d\u3092\u5897\u3084\u3057\u5b66\u7fd2\u7528\u673a\u3092\u5206\u96e2\u3001\u8535\u66f8\u3082\u53e4\u3044\u8cc7\u6599\u3092\u79fb\u7ba1\u3057\u3066\u7d75\u672c\u3084\u5b9f\u7528\u66f8\u3092\u5897\u3084\u3057\u3001\u5efa\u7269\u8a2d\u5099\u3082\u7a7a\u8abf\u3084\u30a8\u30ec\u30d9\u30fc\u30bf\u30fc\u306a\u3069\u3092\u66f4\u65b0\u3057\u305f\u3002\u4e00\u65b9\u3067\u99d0\u8eca\u5834\u3084\u958b\u9928\u6642\u9593\u306e\u8ab2\u984c\u306f\u6b8b\u308a\u3001\u5229\u7528\u8005\u306e\u53cd\u5fdc\u306f\u597d\u610f\u7684\u3060\u304c\u5b66\u7fd2\u5e2d\u306e\u7a7a\u304d\u72b6\u6cc1\u3092\u30db\u30fc\u30e0\u30da\u30fc\u30b8\u3067\u78ba\u8a8d\u3067\u304d\u308b\u4ed5\u7d44\u307f\u3092\u5e74\u5185\u306b\u5c0e\u5165\u3059\u308b\u4e88\u5b9a\u3060\u3002\n<\/code><\/pre>\n<p><code>Q2_K<\/code> 0.94GB<\/p>\n<pre><code class=\"language-text\">\u307f\u3069\u308a\u5e02\u306f\u6625\u306b\u6539\u4fee\u30fb\u518d\u958b\u9928\u3057\u305f\u3002\u6539\u4fee\u306f\u7d04\u4e00\u5e74\u534a\u306b\u308f\u305f\u308a\u3001\u7dcf\u4e8b\u696d\u8cbb\u306f\u7d04\u5341\u4e8c\u5104\u5186\u306e\u3046\u3061\u56fd\u306e\u88dc\u52a9\u91d1\u306f\u7d04\u56db\u5104\u5186\u3068\u306a\u3063\u305f\u3002\u6539\u4fee\u306e\u6700\u5927\u306e\u5909\u5316\u306f\u3001\u95b2\u89a7\u5e2d\u306e\u914d\u7f6e\u3068\u4f1a\u8a71\u30b9\u30da\u30fc\u30b9\u306e\u8a2d\u7f6e\u3002\n\n\u6539\u4fee\u5f8c\u3001\u9928\u5185\u306f\u4f1a\u8a71\u30b9\u30da\u30fc\u30b9\u304c\u4e00\u968e\u306b\u7f6e\u304b\u308c\u3001\u56f3\u66f8\u9928\u306e\u5897\u52a0\u306f\u65b0\u305f\u306a\u5f62\u614b\u3092\u4fdd\u3061\u306a\u304c\u3089\u3001\u96fb\u5b50\u66f8\u7c4d\u306e\u8cb8\u51fa\u30b5\u30fc\u30d3\u30b9\u3082\u59cb\u307e\u3063\u3066\u3044\u308b\u3002\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u3001\u9928\u306e\u6d3b\u7528\u3092\u8003\u3048\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\u76f4\u3057\u3055\u308c\u308b\u3053\u3068\u3068\u306a\u308a\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u306f\u898b\n<\/code><\/pre>\n<p><em>(Cut off at the 1024-token limit.)<\/em><\/p>\n<p><strong>Formal business email<\/strong><\/p>\n<blockquote>\n<p>\u53d6\u5f15\u5148\u306e\u62c5\u5f53\u8005\u306b\u3001\u6765\u9031\u706b\u66dc\u65e5\u306e\u6253\u3061\u5408\u308f\u305b\u3092\u6c34\u66dc\u65e5\u306b\u5909\u66f4\u3057\u3066\u307b\u3057\u3044\u3068\u304a\u9858\u3044\u3059\u308b\u30e1\u30fc\u30eb\u3092\u3001\u4e01\u5be7\u306a\u656c\u8a9e\u3067\u66f8\u3044\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n<\/blockquote>\n<p><em>(Write an email in polite Japanese (keigo) asking a client&#8217;s contact person to move next Tuesday&#8217;s meeting to Wednesday.)<\/em><\/p>\n<p><code>Q8_0<\/code> (baseline) 2.50GB<\/p>\n<pre><code class=\"language-text\">\u4ef6\u540d\uff1a\u304a\u554f\u3044\u5408\u308f\u305b\u306e\u4ef6\n\n\u4ef6\u540d\u306e\u4ef6\u306b\u3064\u304d\u307e\u3057\u3066\u3001\u3054\u9023\u7d61\u3092\u3055\u305b\u3066\u3044\u305f\u3060\u304d\u307e\u3059\u3002\n\n\u3044\u3064\u3082\u304a\u4e16\u8a71\u306b\u306a\u3063\u3066\u304a\u308a\u307e\u3059\u3002\n\n\u6765\u9031\u706b\u66dc\u65e5\u306b\u884c\u308f\u308c\u308b\u6253\u3061\u5408\u308f\u305b\u3092\u3001\u8aa0\u306b\u6050\u308c\u5165\u308a\u307e\u3059\u304c\u3001\u6c34\u66dc\u65e5\u306b\u5909\u66f4\u3057\u3066\u3044\u305f\u3060\u304d\u305f\u304f\u3054\u4f9d\u983c\u7533\u3057\u4e0a\u3052\u307e\u3059\u3002\n\n\u3054\u90fd\u5408\u306e\u826f\u3044\u65e5\u3092\u5fc3\u304c\u3051\u3066\u3001\u3088\u308d\u3057\u304f\u304a\u9858\u3044\u7533\u3057\u4e0a\u3052\u307e\u3059\u3002\n\n\u3088\u308d\u3057\u304f\u304a\u9858\u3044\u3044\u305f\u3057\u307e\u3059\u3002\n<\/code><\/pre>\n<p><code>Q4_K_M<\/code> 1.50GB<\/p>\n<pre><code class=\"language-text\">\u4ef6\u540d\uff1a\u6765\u9031\u306e\u6253\u3061\u5408\u308f\u305b\u65e5\u7a0b\u306e\u5909\u66f4\u306b\u3064\u3044\u3066\n\n\u4ef6\u540d\u306e\u901a\u308a\u3001\u6765\u9031\u706b\u66dc\u65e5\u306b\u884c\u308f\u308c\u308b\u6253\u3061\u5408\u308f\u305b\u3092\u6c34\u66dc\u65e5\u306b\u5909\u66f4\u3057\u3066\u3044\u305f\u3060\u304d\u305f\u3044\u3068\u304a\u9858\u3044\u7533\u3057\u4e0a\u3052\u307e\u3059\u3002\n\n\u3054\u62c5\u5f53\u8005\u69d8\u306b\u306f\u3001\u6539\u3081\u3066\u3054\u9023\u7d61\u3055\u305b\u3066\u3044\u305f\u3060\u304d\u307e\u3059\u3002\n\n\u3054\u4e0d\u4fbf\u3092\u304a\u304b\u3051\u3057\u3066\u3057\u307e\u3044\u7533\u3057\u8a33\u3054\u3056\u3044\u307e\u305b\u3093\u304c\u3001\u4f55\u5352\u3054\u7406\u89e3\u3044\u305f\u3060\u3051\u307e\u3059\u3088\u3046\u304a\u9858\u3044\u7533\u3057\u4e0a\u3052\u307e\u3059\u3002\n\n\u3088\u308d\u3057\u304f\u304a\u9858\u3044\u3044\u305f\u3057\u307e\u3059\u3002\n<\/code><\/pre>\n<p><code>Q2_K<\/code> 0.94GB<\/p>\n<pre><code class=\"language-text\">\u304a\u3055\u308b\uff08\u6c34\u66dc\u65e5\uff09\u306e\u6253\u3061\u5408\u308f\u305b\u3092\u3001\u6765\u9031\uff08\u65e5\u66dc\u65e5\uff09\u306b\u5909\u66f4\u3057\u3066\u3044\u305f\u3060\u304d\u3001\u8b19\u7136\u3068\u304a\u9858\u3044\u3044\u305f\u3057\u307e\u3059\u3002\n\n\u304a\u624b\u539a\u304f\u3001\u8b19\u7136\u3068\u5b58\u3058\u308a\u307e\u3057\u3087\u3046\u3002\n\n\u304a\u5f0f\u306f\u3001\u8b39\u7136\u3068\u5b58\u3058\u308a\u307e\u3057\u3087\u3046\u3002\n\n\u304a\u5f0f\u306f\u3001\u8b39\u7136\u3068\u5b58\u3058\u308a\u307e\u3057\u3087\u3046\u3002\n\n\u304a\u5f0f\u306f\u3001\u8b39\u7136\u3068\u5b58\u3058\u308a\u307e\u3057\u3087\u3046\u3002\n\n\u304a\u5f0f\u306f\u3001\u8b39\u7136\u3068\u5b58\u3058\u308a\u307e\u3057\u3087\u3046\u3002\n\n\u304a\u5f0f\u306f\u3001\u8b39\u7136\u3068\u5b58\u3058\u308a\u307e\u3057\u3087\u3046\u3002\n\n\u304a\u5f0f\u306f\u3001\u8b39\u7136\u3068\u5b58\u3058\u308a\u307e\u3057\u3087\u3046\u3002\n\n\u304a\u5f0f\u306f\u3001\u8b39\u7136\u3068\u5b58\u3058\u308a\u307e\u3057\u3087\u3046\u3002\n\n\u304a\u5f0f\u306f\u3001\u8b39\u7136\u3068\u5b58\u3058\u308a\u307e\u3057\u3087\u3046\u3002\n\n\u304a\u5f0f\u306f\u3001\u8b39\u7136\u3068\u5b58\u3058\u308a\u307e\u3057\u3087\u3046\u3002\n\n\u304a\u5f0f\u306f\u3001\u8b39\u7136\u3068\u5b58\u3058\u308a\u307e\u3057\u3087\u3046\u3002\n\n\u304a\u5f0f\u306f\u3001\u8b39\u7136\u3068\u5b58\u3058\u308a\u307e\u3057\u3087\u3046\u3002\n<\/code><\/pre>\n<p><strong>Word problem<\/strong><\/p>\n<blockquote>\n<p>\u308a\u3093\u3054\u304c1\u500b120\u5186\u3001\u307f\u304b\u3093\u304c1\u500b45\u5186\u3067\u3059\u3002\u308a\u3093\u3054\u30923\u500b\u3068\u307f\u304b\u3093\u30928\u500b\u8cb7\u3063\u3066\u30011,000\u5186\u672d\u3067\u6255\u3044\u307e\u3057\u305f\u3002\u304a\u3064\u308a\u306f\u3044\u304f\u3089\u3067\u3059\u304b\u3002\u8003\u3048\u65b9\u3082\u793a\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n<\/blockquote>\n<p><em>(Apples cost 120 yen each and mandarins cost 45 yen each. You buy 3 apples and 8 mandarins and pay with a 1,000-yen bill. How much change do you get? Show your reasoning.)<\/em><\/p>\n<p><code>Q8_0<\/code> (baseline) 2.50GB<\/p>\n<pre><code class=\"language-text\">\u304a\u624b\u6570\u3042\u308a\u307e\u305b\u3093\u304b\uff1f\n<\/code><\/pre>\n<p><code>Q4_K_M<\/code> 1.50GB<\/p>\n<pre><code class=\"language-text\">\u304a\u624b\u6570\u3067\u3059\u304c\u3001\u8a08\u7b97\u3092\u4e00\u5ea6\u884c\u3063\u3066\u304b\u3089\u7b54\u3048\u3092\u6559\u3048\u3066\u3044\u305f\u3060\u3051\u307e\u3059\u304b\uff1f\n<\/code><\/pre>\n<p><code>Q2_K<\/code> 0.94GB<\/p>\n<pre><code class=\"language-text\">\u3053\u306e\u554f\u984c\u306e\u8a2d\u5b9a\u306f\u6570\u5b66\u7684\u3067\u3001**\u300c\u308a\u3093\u3054\u3068\u307f\u304b\u3093\u306e\u5358\u7d14\u306a\u4fa1\u683c\u5909\u63db\u300d**\u3068\u3044\u3046\u8003\u3048\u65b9\u304c\u3042\u308a\u307e\u3059\u3002\n\n\u307e\u305a\u3001**\u300c\u307f\u304b\u3093\u306e\u5358\u7d14\u306a\u4fa1\u683c\u5909\u63db\u300d**\u3068\u3044\u3046\u8003\u3048\u65b9**\u306f**\u3001**\u300c\u308a\u3093\u3054\u30923\u500b\u3068\u307f\u304b\u3093\u30928\u500b\u8cb7\u3063\u3066\u30011,000\u5186\u672d\u3067\u6255\u3044\u307e\u3057\u305f\u300d**\u3068\u3044\u3046\u6587\u306e **\u300c1,000\u5186\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046 **\u300c1,000\u300d** \u3068\u3044\u3046\n<\/code><\/pre>\n<p><em>(Cut off at the 1024-token limit.)<\/em><\/p>\n<p>Reference: The correct answer is 280 yen (120\u00d73 = 360, 45\u00d78 = 360, 1,000 \u2212 720 = 280).<\/p>\n<p><strong>Knowledge<\/strong><\/p>\n<blockquote>\n<p>\u65e5\u672c\u306e\u90fd\u9053\u5e9c\u770c\u306e\u3046\u3061\u3001\u6d77\u306b\u9762\u3057\u3066\u3044\u306a\u3044\u770c\u3092\u3059\u3079\u3066\u6319\u3052\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n<\/blockquote>\n<p><em>(List all of Japan&#8217;s prefectures that do not border the sea.)<\/em><\/p>\n<p><code>Q8_0<\/code> (baseline) 2.50GB<\/p>\n<pre><code class=\"language-text\">\u65e5\u672c\u306e\u90fd\u9053\u5e9c\u770c\u306e\u3046\u3061\u3001\u6d77\u306b\u9762\u3057\u3066\u3044\u306a\u3044\u770c\u3092\u3059\u3079\u3066\u6319\u3052\u308b\u306e\u306f\u3001\u5730\u7406\u7684\u306a\u5b9a\u7fa9\u3084\u300c\u9762\u7a4d\u306e\u4e00\u90e8\u304c\u6d77\u306b\u63a5\u3059\u308b\u304b\u300d\u3068\u3044\u3046\u57fa\u6e96\u306b\u3088\u3063\u3066\u7b54\u3048\u304c\u7570\u306a\u308b\u305f\u3081\u3001\u53b3\u5bc6\u306b\u7279\u5b9a\u3059\u308b\u3053\u3068\u306f\u3067\u304d\u307e\u305b\u3093\u3002\u4e00\u822c\u7684\u306b\u3001\u6d77\u306b\u9762\u3057\u3066\u3044\u306a\u3044\u770c\uff08\u5185\u9678\u770c\uff09\u306e\u4f8b\u3068\u3057\u3066\u306f\u3001\u5317\u6d77\u9053\uff08\u4e00\u90e8\u306e\u5730\u57df\u306f\u6d77\u306b\u9762\u3059\u308b\u305f\u3081\u9664\u5916\u3055\u308c\u308b\u5834\u5408\u304c\u3042\u308b\uff09\u3001\u9752\u68ee\u770c\u3001\u5ca9\u624b\u770c\u3001\u5bae\u57ce\u770c\u3001\u79cb\u7530\u770c\u3001\u798f\u5cf6\u770c\u3001\u9577\u91ce\u770c\u3001\u5c90\u961c\u770c\u3001\u9759\u5ca1\u770c\u3001\u5c71\u68a8\u770c\u3001\u9577\u91ce\u770c\u3001\u611b\u77e5\u770c\u3001\u5c90\u961c\u770c\u3001\u6ecb\u8cc0\u770c\u3001\u4eac\u90fd\u5e9c\u3001\u5175\u5eab\u770c\u3001\u5948\u826f\u770c\u3001\u548c\u6b4c\u5c71\u770c\u3001\u9ce5\u53d6\u770c\u3001\u5cf6\u6839\u770c\u3001\u5ca1\u5c71\u770c\u3001\u5e83\u5cf6\u770c\u3001\u5c71\u53e3\u770c\u3001\u5fb3\u5cf6\u770c\u3001\u611b\u5a9b\u770c\u3001\u9ad8\u77e5\u770c\u3001\u798f\u5ca1\u770c\u3001\u4f50\u8cc0\u770c\u3001\u9577\u5d0e\u770c\u3001\u718a\u672c\u770c\u3001\u5927\u5206\u770c\u3001\u5bae\u5d0e\u770c\u3001\u9e7f\u5150\u5cf6\u770c\uff08\u4e00\u90e8\u306f\u6d77\u306b\u9762\u3059\u308b\uff09\u306a\u3069\u304c\u3042\u308a\u307e\u3059\u304c\u3001\u3059\u3079\u3066\u306e\u770c\u3092\u4e00\u5217\u306b\u4e26\u3079\u308b\u3053\u3068\u306f\u3067\u304d\u307e\u305b\u3093\u3002\n\n\u3082\u3057\u300c\u6d77\u306b\u9762\u3057\u3066\u3044\u306a\u3044\u770c\u300d\u306e\u5b9a\u7fa9\u3092\u300c\u9762\u7a4d\u306e\u4e00\u90e8\u304c\u6d77\u306b\u63a5\u3059\u308b\u304b\u3069\u3046\u304b\u3067\u533a\u5225\u3059\u308b\u300d\u3068\u3044\u3046\u524d\u63d0\u3067\u3042\u308c\u3070\u3001\u5177\u4f53\u7684\u306a\u30ea\u30b9\u30c8\u3092\u63d0\u4f9b\u3059\u308b\u3053\u3068\u306f\u3067\u304d\u307e\u305b\u3093\u3002\u5730\u7406\u7684\u306a\u60c5\u5831\u306f\u3001\u90fd\u9053\u5e9c\u770c\u306e\u8a73\u7d30\u306a\u5883\u754c\u3092\u78ba\u8a8d\u3059\u308b\u305f\u3081\u306b\u3001\u90fd\u9053\u5e9c\u770c\u5e81\u306e\u516c\u5f0f\u60c5\u5831\u3084\u5730\u56f3\u3092\u53c2\u7167\u3059\u308b\u5fc5\u8981\u304c\u3042\u308a\u307e\u3059\u3002\n<\/code><\/pre>\n<p><code>Q4_K_M<\/code> 1.50GB<\/p>\n<pre><code class=\"language-text\">\u65e5\u672c\u306e\u90fd\u9053\u5e9c\u770c\u306e\u3046\u3061\u3001\u6d77\u306b\u9762\u3057\u3066\u3044\u306a\u3044\u770c\u3092\u3059\u3079\u3066\u6319\u3052\u308b\u3053\u3068\u306f\u3001\u5730\u7406\u7684\u306a\u5b9a\u7fa9\u3084\u300c\u9762\u7a4d\u306e\u7bc4\u56f2\u300d\u306b\u3088\u3063\u3066\u89e3\u91c8\u304c\u5206\u304b\u308c\u308b\u305f\u3081\u3001\u53b3\u5bc6\u306a\u30ea\u30b9\u30c8\u3092\u4e00\u6982\u306b\u6319\u3052\u308b\u306e\u306f\u56f0\u96e3\u3067\u3059\u3002\u4e00\u822c\u7684\u306b\u3001\u6d77\u306b\u9762\u3057\u3066\u3044\u306a\u3044\u770c\uff08\u5185\u9678\u770c\uff09\u306e\u4f8b\u3068\u3057\u3066\u306f\u3001**\u5c71\u5f62\u770c**\u3001**\u9577\u91ce\u770c**\u3001**\u5c90\u961c\u770c**\u3001**\u9759\u5ca1\u770c**\u3001**\u611b\u77e5\u770c**\u3001**\u4e09\u91cd\u770c**\u3001**\u6ecb\u8cc0\u770c**\u3001**\u4eac\u90fd\u5e9c**\u3001**\u5927\u962a\u5e9c**\u3001**\u5175\u5eab\u770c**\u3001**\u5948\u826f\u770c**\u3001**\u548c\u6b4c\u5c71\u770c**\u3001**\u9ce5\u53d6\u770c**\u3001**\u5cf6\u6839\u770c**\u3001**\u5ca1\u5c71\u770c**\u3001**\u5e83\u5cf6\u770c**\u3001**\u5c71\u53e3\u770c**\u3001**\u5fb3\u5cf6\u770c**\u3001**\u9999\u5ddd\u770c**\u3001**\u611b\u5a9b\u770c**\u3001**\u9ad8\u77e5\u770c**\u3001**\u798f\u5ca1\u770c**\u3001**\u4f50\u8cc0\u770c**\u3001**\u9577\u5d0e\u770c**\u3001**\u718a\u672c\u770c**\u3001**\u5927\u5206\u770c**\u3001**\u5bae\u5d0e\u770c**\u3001**\u9e7f\u5150\u5cf6\u770c**\u3001**\u6c96\u7e04\u770c**\uff08\u96e2\u5cf6\uff09\u306a\u3069\u304c\u6319\u3052\u3089\u308c\u307e\u3059\u3002\n\n\u305f\u3060\u3057\u3001**\u6c96\u7e04\u770c**\u306f\u6d77\u306b\u9762\u3057\u3066\u3044\u308b\u305f\u3081\u9664\u5916\u3055\u308c\u3001**\u5c71\u5f62\u770c**\u306f\u5317\u9678\u5730\u65b9\u306e\u4e00\u90e8\u3067\u6d77\u306b\u9762\u3057\u3066\u3044\u308b\u53ef\u80fd\u6027\u3082\u3042\u308a\u307e\u3059\uff08\u8a73\u7d30\u306a\u5730\u7406\u7684\u5883\u754c\u3092\u78ba\u8a8d\u3059\u308b\u5fc5\u8981\u304c\u3042\u308a\u307e\u3059\uff09\u3002\u307e\u305f\u3001**\u5317\u6d77\u9053**\u3084**\u9752\u68ee\u770c**\u306a\u3069\u306f\u6cbf\u5cb8\u90e8\u306b\u9762\u3057\u3066\u3044\u308b\u305f\u3081\u3001\u5185\u9678\u770c\u3068\u306f\u307f\u306a\u3055\u308c\u307e\u305b\u3093\u3002\n\n\u3082\u3057\u300c\u6d77\u306b\u9762\u3057\u3066\u3044\u306a\u3044\u770c\u300d\u306e\u5b9a\u7fa9\u3092\u300c\u6cbf\u5cb8\u90e8\uff08\u6d77\u306b\u63a5\u3059\u308b\u5730\u57df\uff09\u3067\u306f\u306a\u3044\u770c\u300d\u3068\u7406\u89e3\u3059\u308b\u306a\u3089\u3001**\u5c71\u5f62\u770c**\u3001**\u9577\u91ce\u770c**\u3001**\u5c90\u961c\u770c**\u3001**\u9759\u5ca1\u770c**\u3001**\u611b\u77e5\u770c**\u3001**\u4e09\u91cd\u770c**\u3001**\u6ecb\u8cc0\u770c**\u3001**\u4eac\u90fd\u5e9c**\u3001**\u5927\u962a\u5e9c**\u3001**\u5175\u5eab\u770c**\u3001**\u5948\u826f\u770c**\u3001**\u548c\u6b4c\u5c71\u770c**\u3001**\u9ce5\u53d6\u770c**\u3001**\u5cf6\u6839\u770c**\u3001**\u5ca1\u5c71\u770c**\u3001**\u5e83\u5cf6\u770c**\u3001**\u5c71\u53e3\u770c**\u3001**\u5fb3\u5cf6\u770c**\u3001**\u9999\u5ddd\u770c**\u3001**\u611b\u5a9b\u770c**\u3001**\u9ad8\u77e5\u770c**\u3001**\u798f\u5ca1\u770c**\u3001**\u4f50\u8cc0\u770c**\u3001**\u9577\u5d0e\u770c**\u3001**\u718a\u672c\u770c**\u3001**\u5927\u5206\u770c**\u3001**\u5bae\u5d0e\u770c**\u3001**\u9e7f\u5150\u5cf6\u770c**\u3001**\u5317\u6d77\u9053**\u3001**\u9752\u68ee\u770c**\u306a\u3069\u304c\u8a72\u5f53\u3057\u307e\u3059\u3002\n\n\u5177\u4f53\u7684\u306a\u5b9a\u7fa9\u3084\u57fa\u6e96\u306b\u3088\u3063\u3066\u7570\u306a\u308b\u305f\u3081\u3001\u3088\u308a\u6b63\u78ba\u306a\u60c5\u5831\u306f\u65e5\u672c\u306e\u5730\u7406\u306e\u8cc7\u6599\u3092\u53c2\u7167\u3059\u308b\u5fc5\u8981\u304c\u3042\u308a\u307e\u3059\u3002\n<\/code><\/pre>\n<p><code>Q2_K<\/code> 0.94GB<\/p>\n<pre><code class=\"language-text\">\u65e5\u672c\u306e\u90fd\u9053\u5e9c\u770c\u306f\u3059\u3079\u3066\u6d77\u306b\u9762\u3057\u3066\u3044\u306a\u3044\u308f\u3051\u3067\u306f\u306a\u304f\u3001\u3059\u3079\u3066\u306e\u770c\u304c\u6d77\u306b\u9762\u3057\u3066\u3044\u308b\u308f\u3051\u3067\u306f\u3042\u308a\u307e\u305b\u3093\u3002\u5177\u4f53\u7684\u306a\u56de\u7b54\u306f\u5b58\u5728\u3057\u307e\u305b\u3093\u304c\u3001\u4e00\u822c\u7684\u306a\u77e5\u8b58\u3068\u3057\u3066\u3001\u6c96\u7e21\u770c\u306f\u6d77\u306b\u9762\u3057\u3066\u3044\u307e\u3059\u3002\u4ed6\u306e\u770c\u306f\u6d77\u306b\u9762\u3057\u3066\u3044\u306a\u3044\u304b\u3069\u3046\u304b\u306f\u3001\u5177\u4f53\u7684\u306a\u60c5\u5831\u304c\u306a\u3044\u305f\u3081\u3001\u56de\u7b54\u306f\u3067\u304d\u307e\u305b\u3093\u3002\n<\/code><\/pre>\n<p>Reference: The correct answer is eight prefectures: Tochigi, Gunma, Saitama, Yamanashi, Nagano, Gifu, Shiga and Nara.<\/p>\n<p><strong>Constrained writing<\/strong><\/p>\n<blockquote>\n<p>\u300c\u6625\u300d\u3092\u30c6\u30fc\u30de\u306b\u3057\u305f\u4ff3\u53e5\u30923\u3064\u3001\u4e94\u30fb\u4e03\u30fb\u4e94\u306e\u97f3\u6570\u3092\u5b88\u3063\u3066\u4f5c\u3063\u3066\u304f\u3060\u3055\u3044\u3002\u305d\u308c\u305e\u308c\u306b\u4e00\u8a00\u306e\u8aac\u660e\u3092\u4ed8\u3051\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n<\/blockquote>\n<p><em>(Write three haiku on the theme of spring, keeping the 5-7-5 syllable pattern, each with a one-line explanation.)<\/em><\/p>\n<p><code>Q8_0<\/code> (baseline) 2.50GB<\/p>\n<pre><code class=\"language-text\">\u6625\u306e\u98a8\u304c\u5439\u304f\u3002\n\u7a7a\u3092\u5e83\u304f\u611f\u3058\u308b\u3002\n\n\u6625\u306e\u82b1\u304c\u54b2\u304f\u3002\n\u9759\u3051\u3055\u304c\u5897\u3059\u3002\n\n\u6625\u306e\u6c34\u304c\u6e05\u304f\u3002\n\u5fc3\u3082\u6f84\u307e\u3059\u3002\n<\/code><\/pre>\n<p><code>Q4_K_M<\/code> 1.50GB<\/p>\n<pre><code class=\"language-text\">\u6625\u306f\u3001\u82bd\u5439\u304f\u7a7a\u3092\u547c\u3073\u307e\u3059\u3002\n\uff08\u65b0\u305f\u306a\u5e0c\u671b\u3092\u8c61\u5fb4\u3059\u308b\uff09\n\n\u6625\u306f\u3001\u98a8\u304c\u67d4\u3089\u304b\u304f\u5439\u304f\u3002\n\uff08\u5fc3\u5730\u3088\u3044\u6c17\u5206\u3092\u8868\u3059\uff09\n\n\u6625\u306f\u3001\u7dd1\u304c\u5e83\u304c\u308b\u9803\u3002\n\uff08\u81ea\u7136\u306e\u7f8e\u3092\u611b\u3059\u308b\uff09\n<\/code><\/pre>\n<p><code>Q2_K<\/code> 0.94GB<\/p>\n<pre><code class=\"language-text\">**\u6625**\n\n**\u6625**\n\u6625\u306f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\u304f\u9060\n<\/code><\/pre>\n<p><em>(Cut off at the 1024-token limit.)<\/em><\/p>\n<h3>Running It on a CPU Only<\/h3>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Item<\/th>\n<th>Result<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Prompt processing (512 tokens)<\/td>\n<td>34.4 tokens\/s<\/td>\n<\/tr>\n<tr>\n<td>Generation (128 tokens)<\/td>\n<td>13.9 tokens\/s<\/td>\n<\/tr>\n<tr>\n<td>Generation in Japanese characters<\/td>\n<td>about 20.9 characters\/s<\/td>\n<\/tr>\n<tr>\n<td>Peak memory (context 4,096 tokens)<\/td>\n<td>2.9GB<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p>Conditions: Neoverse-N1, 3 threads, no GPU, llama.cpp b11223, <code>Q4_K_M<\/code> (1.45GB) from <a href=\"https:\/\/huggingface.co\/openbmb\/MiniCPM5-2B-GGUF\">openbmb\/MiniCPM5-2B-GGUF<\/a>, measured 2026-09-28 (JST). Other models measured under the same conditions (closest in file size):<\/p>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<p class=\"lmw-table-hint\" style=\"margin:0 0 4px;font-size:0.85em;opacity:0.7;\">\u2192 Scroll horizontally to see all columns<\/p>\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Model<\/th>\n<th>Quant<\/th>\n<th>File size<\/th>\n<th>Generation<\/th>\n<th>In Japanese<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>Qwen3-4B<\/td>\n<td><code>Q4_K_M<\/code><\/td>\n<td>2.33GB<\/td>\n<td>6.8 tokens\/s<\/td>\n<td>\u2014<\/td>\n<\/tr>\n<tr>\n<td><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/18\/tencent-releases-wevisdoc-document-parsing-models\/\">WeVisDoc-4B<\/a><\/td>\n<td><code>Q4_K_M<\/code><\/td>\n<td>2.53GB<\/td>\n<td>8.3 tokens\/s<\/td>\n<td>12.1 chars\/s<\/td>\n<\/tr>\n<tr>\n<td><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/16\/eleutherai-olmo-3-7b-reward-hacking-models\/\">olmo3-7b-sdf-sft-clean150<\/a><\/td>\n<td><code>Q4_K_M<\/code><\/td>\n<td>4.16GB<\/td>\n<td>4.9 tokens\/s<\/td>\n<td>4.4 chars\/s<\/td>\n<\/tr>\n<tr>\n<td>Qwen3-8B<\/td>\n<td><code>Q4_K_M<\/code><\/td>\n<td>4.68GB<\/td>\n<td>4.6 tokens\/s<\/td>\n<td>\u2014<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p>Peak memory includes the model file mapped into memory (mmap). A PC with a GPU will be much faster; these figures are a guide for CPU-only machines such as mini PCs and ARM boards.<\/p>\n<h3>Answers to Our Japanese Questions (Verbatim)<\/h3>\n<p>We ask every model the same five questions in Japanese under the conditions above (temperature 0, thinking turned off where the chat template allows it, up to 1024 tokens) and publish the answers unedited. We do not grade them; where a question has a single correct answer, we show it for comparison.<\/p>\n<p><strong>Summarization<\/strong><\/p>\n<blockquote>\n<p>\u6b21\u306e\u8a18\u4e8b\u30923\u6587\u3067\u8981\u7d04\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n<p>\u307f\u3069\u308a\u5e02\u306f\u4eca\u5e74\u306e\u6625\u3001\u5e02\u5185\u306b\u4e09\u3064\u3042\u308b\u516c\u7acb\u56f3\u66f8\u9928\u306e\u3046\u3061\u6700\u3082\u53e4\u3044\u4e2d\u592e\u56f3\u66f8\u9928\u306e\u6539\u4fee\u3092\u7d42\u3048\u3001\u5341\u6708\u4e00\u65e5\u306b\u518d\u958b\u9928\u3057\u305f\u3002\u6539\u4fee\u306f\u7d04\u4e00\u5e74\u534a\u306b\u308f\u305f\u308a\u3001\u7dcf\u4e8b\u696d\u8cbb\u306f\u7d04\u5341\u4e8c\u5104\u5186\u3002\u3053\u306e\u3046\u3061\u7d04\u56db\u5104\u5186\u306f\u56fd\u306e\u88dc\u52a9\u91d1\u3067\u307e\u304b\u306a\u308f\u308c\u305f\u3002<\/p>\n<p>\u6700\u5927\u306e\u5909\u5316\u306f\u3001\u95b2\u89a7\u5e2d\u306e\u914d\u7f6e\u3067\u3042\u308b\u3002\u4ee5\u524d\u306f\u9928\u5185\u306e\u5ea7\u5e2d\u306e\u5927\u534a\u304c\u5b66\u7fd2\u7528\u306e\u673a\u3067\u3001\u8a66\u9a13\u671f\u9593\u306b\u306a\u308b\u3068\u4e2d\u9ad8\u751f\u3067\u57cb\u307e\u308a\u3001\u9ad8\u9f62\u306e\u5229\u7528\u8005\u3084\u5c0f\u3055\u306a\u5b50\u3069\u3082\u9023\u308c\u304c\u5ea7\u308c\u306a\u3044\u3068\u3044\u3046\u82e6\u60c5\u304c\u6bce\u5e74\u5bc4\u305b\u3089\u308c\u3066\u3044\u305f\u3002\u6539\u4fee\u5f8c\u306f\u5ea7\u5e2d\u6570\u3092\u767e\u4e8c\u5341\u5e2d\u304b\u3089\u4e8c\u767e\u5e2d\u306b\u5897\u3084\u3057\u305f\u3046\u3048\u3067\u3001\u5b66\u7fd2\u7528\u306e\u673a\u3092\u4e09\u968e\u306b\u307e\u3068\u3081\u3001\u4e00\u968e\u306f\u4f1a\u8a71\u304c\u3067\u304d\u308b\u300c\u306b\u304e\u308f\u3044\u30d5\u30ed\u30a2\u300d\u3001\u4e8c\u968e\u306f\u9759\u304b\u306b\u672c\u3092\u8aad\u3080\u300c\u95b2\u89a7\u30d5\u30ed\u30a2\u300d\u306b\u5206\u3051\u305f\u3002\u4e00\u968e\u306b\u306f\u4e88\u7d04\u5236\u306e\u30b0\u30eb\u30fc\u30d7\u5b66\u7fd2\u5ba4\u304c\u56db\u5ba4\u3068\u3001\u98f2\u307f\u7269\u3092\u6301\u3061\u8fbc\u3081\u308b\u30ab\u30d5\u30a7\u30b9\u30da\u30fc\u30b9\u3082\u8a2d\u3051\u3089\u308c\u305f\u3002<\/p>\n<p>\u8535\u66f8\u306e\u9762\u3067\u3082\u898b\u76f4\u3057\u304c\u9032\u3093\u3060\u3002\u8cb8\u51fa\u306e\u5c11\u306a\u3044\u53e4\u3044\u8cc7\u6599\u7d04\u4e09\u4e07\u518a\u3092\u5e02\u5185\u306e\u4ed6\u306e\u9928\u3084\u4fdd\u5b58\u66f8\u5eab\u306b\u79fb\u3057\u3001\u4ee3\u308f\u308a\u306b\u5b50\u3069\u3082\u5411\u3051\u306e\u7d75\u672c\u3068\u3001\u4ecb\u8b77\u30fb\u5b50\u80b2\u3066\u30fb\u4ed5\u4e8b\u63a2\u3057\u3068\u3044\u3063\u305f\u66ae\u3089\u3057\u306b\u95a2\u308f\u308b\u5b9f\u7528\u66f8\u3092\u7d04\u4e00\u4e07\u4e94\u5343\u518a\u5897\u3084\u3057\u305f\u3002\u96fb\u5b50\u66f8\u7c4d\u306e\u8cb8\u51fa\u30b5\u30fc\u30d3\u30b9\u3082\u59cb\u307e\u308a\u3001\u5229\u7528\u767b\u9332\u3092\u3059\u308c\u3070\u81ea\u5b85\u304b\u3089\u4e00\u5ea6\u306b\u4e09\u518a\u307e\u3067\u3001\u4e8c\u9031\u9593\u501f\u308a\u3089\u308c\u308b\u3002<\/p>\n<p>\u5efa\u7269\u305d\u306e\u3082\u306e\u306b\u3082\u624b\u304c\u5165\u3063\u305f\u3002\u7bc9\u56db\u5341\u5e74\u3092\u8d85\u3048\u3066\u3044\u305f\u7a7a\u8abf\u8a2d\u5099\u3092\u5165\u308c\u66ff\u3048\u3001\u5c4b\u4e0a\u306b\u306f\u592a\u967d\u5149\u30d1\u30cd\u30eb\u3092\u8a2d\u7f6e\u3057\u3066\u3001\u9928\u5185\u3067\u4f7f\u3046\u96fb\u529b\u306e\u7d04\u4e8c\u5272\u3092\u307e\u304b\u306a\u3048\u308b\u3088\u3046\u306b\u3057\u305f\u3002\u8eca\u3044\u3059\u306e\u5229\u7528\u8005\u304b\u3089\u8981\u671b\u306e\u591a\u304b\u3063\u305f\u30a8\u30ec\u30d9\u30fc\u30bf\u30fc\u306f\u4e00\u57fa\u304b\u3089\u4e8c\u57fa\u306b\u5897\u3048\u3001\u66f8\u67b6\u306e\u9593\u306e\u901a\u8def\u3082\u4ee5\u524d\u3088\u308a\u4e09\u5341\u30bb\u30f3\u30c1\u5e83\u3052\u3089\u308c\u305f\u3002\u5165\u53e3\u306b\u306f\u6bb5\u5dee\u306e\u306a\u3044\u30b9\u30ed\u30fc\u30d7\u304c\u65b0\u305f\u306b\u8a2d\u3051\u3089\u308c\u3001\u6388\u4e73\u5ba4\u3068\u591a\u76ee\u7684\u30c8\u30a4\u30ec\u3082\u5404\u968e\u306b\u4e00\u3064\u305a\u3064\u7f6e\u304b\u308c\u305f\u3002<\/p>\n<p>\u4e00\u65b9\u3067\u8ab2\u984c\u3082\u6b8b\u308b\u3002\u99d0\u8eca\u5834\u306f\u6539\u4fee\u524d\u3068\u540c\u3058\u56db\u5341\u53f0\u306e\u307e\u307e\u3067\u3001\u9031\u672b\u306b\u306f\u8fd1\u304f\u306e\u9053\u8def\u306b\u9806\u756a\u5f85\u3061\u306e\u8eca\u304c\u4e26\u3093\u3060\u3002\u5e02\u306f\u96a3\u63a5\u3059\u308b\u5e02\u6709\u5730\u3092\u81e8\u6642\u99d0\u8eca\u5834\u3068\u3057\u3066\u958b\u653e\u3057\u305f\u304c\u3001\u6052\u4e45\u7684\u306a\u5bfe\u7b56\u306f\u6c7a\u307e\u3063\u3066\u3044\u306a\u3044\u3002\u307e\u305f\u3001\u958b\u9928\u6642\u9593\u3092\u5e73\u65e5\u306e\u591c\u4e5d\u6642\u307e\u3067\u5ef6\u3070\u3057\u305f\u3053\u3068\u3067\u8077\u54e1\u306e\u8ca0\u62c5\u304c\u5897\u3048\u3066\u304a\u308a\u3001\u8077\u54e1\u56e3\u4f53\u306f\u4eba\u54e1\u306e\u8ffd\u52a0\u3092\u6c42\u3081\u3066\u3044\u308b\u3002<\/p>\n<p>\u5229\u7528\u8005\u306e\u53d7\u3051\u6b62\u3081\u306f\u304a\u304a\u3080\u306d\u597d\u610f\u7684\u3060\u3002\u9031\u306b\u4e8c\u5ea6\u3001\u5b6b\u3092\u9023\u308c\u3066\u6765\u308b\u3068\u3044\u3046\u4e03\u5341\u4ee3\u306e\u5973\u6027\u306f\u300c\u524d\u306f\u5b50\u3069\u3082\u304c\u58f0\u3092\u51fa\u3059\u3068\u5468\u308a\u306e\u76ee\u304c\u6c17\u306b\u306a\u3063\u305f\u304c\u3001\u4e00\u968e\u306a\u3089\u6c17\u517c\u306d\u306a\u304f\u7d75\u672c\u3092\u8aad\u3093\u3067\u3042\u3052\u3089\u308c\u308b\u300d\u3068\u8a71\u3059\u3002\u4e00\u65b9\u3001\u53d7\u9a13\u3092\u63a7\u3048\u305f\u9ad8\u6821\u751f\u306e\u4e2d\u306b\u306f\u300c\u5b66\u7fd2\u5e2d\u304c\u4e09\u968e\u3060\u3051\u306b\u306a\u308a\u3001\u5915\u65b9\u306f\u3059\u3050\u306b\u6e80\u5e2d\u306b\u306a\u308b\u300d\u3068\u4e0d\u6e80\u3092\u6f0f\u3089\u3059\u4eba\u3082\u3044\u308b\u3002\u56f3\u66f8\u9928\u306f\u3001\u5b66\u7fd2\u5e2d\u306e\u7a7a\u304d\u72b6\u6cc1\u3092\u30db\u30fc\u30e0\u30da\u30fc\u30b8\u3067\u78ba\u8a8d\u3067\u304d\u308b\u4ed5\u7d44\u307f\u3092\u5e74\u5185\u306b\u5c0e\u5165\u3059\u308b\u4e88\u5b9a\u3060\u3002<\/p>\n<p>\u518d\u958b\u9928\u306b\u5408\u308f\u305b\u3066\u3001\u50ac\u3057\u3082\u5897\u3048\u305f\u3002\u6bce\u9031\u571f\u66dc\u306e\u5348\u524d\u306b\u306f\u8aad\u307f\u805e\u304b\u305b\u4f1a\u3001\u6708\u306b\u4e00\u5ea6\u306f\u5730\u5143\u306e\u8fb2\u5bb6\u3084\u5546\u5e97\u4e3b\u3092\u8b1b\u5e2b\u306b\u62db\u304f\u300c\u307e\u3061\u306e\u5148\u751f\u300d\u8b1b\u5ea7\u304c\u958b\u304b\u308c\u308b\u3002\u5341\u4e00\u6708\u306b\u306f\u3001\u5e02\u5185\u306e\u4e2d\u5b66\u751f\u304c\u9078\u3093\u3060\u672c\u3092\u7d39\u4ecb\u3059\u308b\u5c55\u793a\u3082\u4e88\u5b9a\u3055\u308c\u3066\u3044\u308b\u3002<\/p>\n<p>\u9928\u9577\u306e\u5c71\u672c\u548c\u5b50\u3055\u3093\u306f\u300c\u672c\u3092\u501f\u308a\u308b\u5834\u6240\u304b\u3089\u3001\u4eba\u304c\u96c6\u307e\u3063\u3066\u904e\u3054\u3059\u5834\u6240\u3078\u5909\u3048\u305f\u304b\u3063\u305f\u3002\u518d\u958b\u9928\u304b\u3089\u306e\u4e00\u304b\u6708\u3067\u6765\u9928\u8005\u306f\u4ee5\u524d\u306e\u4e00\u30fb\u516b\u500d\u306b\u306a\u3063\u305f\u300d\u3068\u8a71\u3059\u3002\u5e02\u306f\u6765\u5e74\u5ea6\u3001\u6b8b\u308b\u4e8c\u3064\u306e\u56f3\u66f8\u9928\u306b\u3064\u3044\u3066\u3082\u5229\u7528\u8005\u3078\u306e\u805e\u304d\u53d6\u308a\u3092\u884c\u3044\u3001\u6539\u4fee\u306e\u5fc5\u8981\u6027\u3092\u691c\u8a0e\u3059\u308b\u65b9\u91dd\u3060\u3002<\/p>\n<\/blockquote>\n<p><em>(Summarize the following article in three sentences. (The article, about 1,200 Japanese characters, describes the renovation and reopening of a city&#8217;s central library: the budget, the new floor layout and seating, changes to the collection and e-book lending, building and accessibility work, remaining issues such as parking and staff workload, users&#8217; reactions, new events, and the city&#8217;s next steps.))<\/em><\/p>\n<pre><code class=\"language-text\">\u307f\u3069\u308a\u5e02\u306f\u6625\u306b\u4e2d\u592e\u56f3\u66f8\u9928\u3092\u6539\u4fee\u518d\u958b\u3057\u3001\u7dcf\u4e8b\u696d\u8cbb\u7d04\u5341\u4e8c\u5104\u5186\uff08\u7d04\u56db\u5104\u5186\u306f\u56fd\u306e\u88dc\u52a9\u91d1\uff09\u3092\u6295\u5165\u3057\u307e\u3057\u305f\u3002\u6539\u4fee\u306b\u3088\u308a\u5ea7\u5e2d\u6570\u3092\u5897\u3084\u3057\u3001\u6a5f\u68b0\u5ba4\u3092\u5206\u985e\u3057\u3001\u8535\u66f8\u3082\u53e4\u3044\u8cc7\u6599\u304b\u3089\u5b9f\u7528\u66f8\u3078\u3068\u898b\u76f4\u3057\u307e\u3057\u305f\u3002\u307e\u305f\u3001\u7a7a\u8abf\u3084\u30a8\u30ec\u30d9\u30fc\u30bf\u30fc\u306a\u3069\u5efa\u7269\u81ea\u4f53\u3082\u66f4\u65b0\u3055\u308c\u307e\u3057\u305f\u304c\u3001\u99d0\u8eca\u5834\u3084\u958b\u9928\u6642\u9593\u306e\u8ab2\u984c\u3082\u6b8b\u3063\u3066\u3044\u307e\u3059\u3002\u9928\u9577\u306f\u6765\u9928\u8005\u6570\u304c\u4e00\u304b\u6708\u3067\u4e00\u516b\u500d\u306b\u306a\u3063\u305f\u3068\u8ff0\u3079\u3001\u6b8b\u308a\u306e\u56f3\u66f8\u9928\u3082\u6539\u4fee\u3092\u691c\u8a0e\u3059\u308b\u65b9\u91dd\u3067\u3059\u3002\n<\/code><\/pre>\n<p><strong>Formal business email<\/strong><\/p>\n<blockquote>\n<p>\u53d6\u5f15\u5148\u306e\u62c5\u5f53\u8005\u306b\u3001\u6765\u9031\u706b\u66dc\u65e5\u306e\u6253\u3061\u5408\u308f\u305b\u3092\u6c34\u66dc\u65e5\u306b\u5909\u66f4\u3057\u3066\u307b\u3057\u3044\u3068\u304a\u9858\u3044\u3059\u308b\u30e1\u30fc\u30eb\u3092\u3001\u4e01\u5be7\u306a\u656c\u8a9e\u3067\u66f8\u3044\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n<\/blockquote>\n<p><em>(Write an email in polite Japanese (keigo) asking a client&#8217;s contact person to move next Tuesday&#8217;s meeting to Wednesday.)<\/em><\/p>\n<pre><code class=\"language-text\">\u4ef6\u540d\uff1a\u6765\u9031\u306e\u6253\u3061\u5408\u308f\u305b\u65e5\u7a0b\u306e\u3054\u8abf\u6574\u306b\u3064\u3044\u3066\n\n\u4ef6\u540d\uff1a\u6765\u9031\u306e\u6253\u3061\u5408\u308f\u305b\u65e5\u7a0b\u306e\u3054\u8abf\u6574\u306b\u3064\u3044\u3066\n\n\u3044\u3064\u3082\u304a\u4e16\u8a71\u306b\u306a\u3063\u3066\u304a\u308a\u307e\u3059\u3002\n\u3007\u3007\uff08\u540d\u524d\uff09\u3067\u3059\u3002\n\n\u3044\u3064\u3082\u304a\u4e16\u8a71\u306b\u306a\u3063\u3066\u304a\u308a\u307e\u3059\u3002\n\u3007\u3007\uff08\u540d\u524d\uff09\u3067\u3059\u3002\n\n\u6765\u9031\u706b\u66dc\u65e5\u306b\u884c\u308f\u308c\u308b\u6253\u3061\u5408\u308f\u305b\u3092\u3001\u6c34\u66dc\u65e5\u306b\u304a\u9858\u3044\u3044\u305f\u3057\u307e\u3059\u3002\n\u304a\u624b\u4f1d\u3044\u3067\u304d\u308b\u3088\u3046\u3001\u5f15\u304d\u7d9a\u304d\u3088\u308d\u3057\u304f\u304a\u9858\u3044\u3044\u305f\u3057\u307e\u3059\u3002\n\n\u3007\u3007\uff08\u540d\u524d\uff09\n<\/code><\/pre>\n<p><strong>Word problem<\/strong><\/p>\n<blockquote>\n<p>\u308a\u3093\u3054\u304c1\u500b120\u5186\u3001\u307f\u304b\u3093\u304c1\u500b45\u5186\u3067\u3059\u3002\u308a\u3093\u3054\u30923\u500b\u3068\u307f\u304b\u3093\u30928\u500b\u8cb7\u3063\u3066\u30011,000\u5186\u672d\u3067\u6255\u3044\u307e\u3057\u305f\u3002\u304a\u3064\u308a\u306f\u3044\u304f\u3089\u3067\u3059\u304b\u3002\u8003\u3048\u65b9\u3082\u793a\u3057\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n<\/blockquote>\n<p><em>(Apples cost 120 yen each and mandarins cost 45 yen each. You buy 3 apples and 8 mandarins and pay with a 1,000-yen bill. How much change do you get? Show your reasoning.)<\/em><\/p>\n<pre><code class=\"language-text\">\u307e\u305a\u3001\u308a\u3093\u3054\u306e\u5408\u8a08\u91d1\u984d\u3068\u307f\u304b\u3093\u306e\u5408\u8a08\u91d1\u984d\u3092\u8a08\u7b97\u3057\u307e\u3059\u3002\n\u308a\u3093\u3054\u306f1\u500b120\u5186\u306a\u306e\u3067\u30013\u500b\u3060\u3068120\u5186\u00d73=360\u5186\u306b\u306a\u308a\u307e\u3059\u3002\n\u307f\u304b\u3093\u306f1\u500b45\u5186\u306a\u306e\u3067\u30018\u500b\u3060\u306845\u5186\u00d78=360\u5186\u306b\u306a\u308a\u307e\u3059\u3002\n\n\u6b21\u306b\u3001\u3053\u308c\u3089\u3092\u5408\u8a08\u3057\u307e\u3059\u3002\n\u308a\u3093\u3054\u3068\u307f\u304b\u3093\u306e\u5408\u8a08\u91d1\u984d\u306f360\u5186+360\u5186=720\u5186\u306b\u306a\u308a\u307e\u3059\u3002\n\n\u6700\u5f8c\u306b\u3001\u652f\u6255\u3063\u305f\u91d1\u984d\u304b\u3089\u5408\u8a08\u91d1\u984d\u3092\u5f15\u3044\u3066\u304a\u3064\u308a\u3092\u8a08\u7b97\u3057\u307e\u3059\u3002\n1,000\u5186\u304b\u3089720\u5186\u3092\u5f15\u304f\u3068\u30011,000\u5186-720\u5186=280\u5186\u306b\u306a\u308a\u307e\u3059\u3002\n\n\u304a\u3064\u308a\u306f280\u5186\u3067\u3059\u3002\n<\/code><\/pre>\n<p>Reference: The correct answer is 280 yen (120\u00d73 = 360, 45\u00d78 = 360, 1,000 \u2212 720 = 280).<\/p>\n<p><strong>Knowledge<\/strong><\/p>\n<blockquote>\n<p>\u65e5\u672c\u306e\u90fd\u9053\u5e9c\u770c\u306e\u3046\u3061\u3001\u6d77\u306b\u9762\u3057\u3066\u3044\u306a\u3044\u770c\u3092\u3059\u3079\u3066\u6319\u3052\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n<\/blockquote>\n<p><em>(List all of Japan&#8217;s prefectures that do not border the sea.)<\/em><\/p>\n<pre><code class=\"language-text\">\u65e5\u672c\u306e\u90fd\u9053\u5e9c\u770c\u306e\u3046\u3061\u3001\u6d77\u306b\u9762\u3057\u3066\u3044\u306a\u3044\u770c\u3092\u3059\u3079\u3066\u6319\u3052\u308b\u3068\u3001\u4ee5\u4e0b\u306e\u770c\u304c\u3042\u308a\u307e\u3059\u3002\n\n1. \u9752\u68ee\u770c\n2. \u5ca9\u624b\u770c\n3. \u5bae\u57ce\u770c\n4. \u79cb\u7530\u770c\n5. \u5c71\u5f62\u770c\n6. \u798f\u5cf6\u770c\n7. \u8328\u57ce\u770c\n8. \u6803\u6728\u770c\n9. \u7fa4\u99ac\u770c\n10. \u57fc\u7389\u770c\n11. \u5343\u8449\u770c\n12. \u6771\u4eac\u90fd\n13. \u795e\u5948\u5ddd\u770c\n14. \u65b0\u6f5f\u770c\n15. \u5bcc\u5c71\u770c\n16. \u77f3\u5ddd\u770c\n17. \u798f\u4e95\u770c\n18. \u5c71\u68a8\u770c\n19. \u9577\u91ce\u770c\n20. \u5c90\u961c\u770c\n21. \u9759\u5ca1\u770c\n22. \u611b\u77e5\u770c\n23. \u4e09\u91cd\u770c\n24. \u6ecb\u8cc0\u770c\n25. \u4eac\u90fd\u5e9c\n26. \u5175\u5eab\u770c\n27. \u5948\u826f\u770c\n28. \u548c\u6b4c\u5c71\u770c\n29. \u9ce5\u53d6\u770c\n30. \u5cf6\u6839\u770c\n31. \u5ca1\u5c71\u770c\n32. \u5e83\u5cf6\u770c\n33. \u5c71\u53e3\u770c\n34. \u5fb3\u5cf6\u770c\n35. \u9999\u5ddd\u770c\n36. \u611b\u5a9b\u770c\n37. \u9ad8\u77e5\u770c\n38. \u798f\u5ca1\u770c\n39. \u4f50\u8cc0\u770c\n40. \u9577\u5d0e\u770c\n41. \u718a\u672c\u770c\n42. \u5927\u5206\u770c\n43. \u5bae\u5d0e\u770c\n44. \u9e7f\u5150\u5cf6\u770c\n45. \u6c96\u7e04\u770c\n\n\u306a\u304a\u3001\u6c96\u7e04\u770c\u306f\u592a\u5e73\u6d0b\u306b\u9762\u3057\u3066\u3044\u307e\u3059\u304c\u3001\u56fd\u5185\u306e\u4e3b\u8981\u306a\u6d77\u306b\u9762\u3057\u3066\u3044\u306a\u3044\u770c\u3067\u3059\u3002\n<\/code><\/pre>\n<p>Reference: The correct answer is eight prefectures: Tochigi, Gunma, Saitama, Yamanashi, Nagano, Gifu, Shiga and Nara.<\/p>\n<p><strong>Constrained writing<\/strong><\/p>\n<blockquote>\n<p>\u300c\u6625\u300d\u3092\u30c6\u30fc\u30de\u306b\u3057\u305f\u4ff3\u53e5\u30923\u3064\u3001\u4e94\u30fb\u4e03\u30fb\u4e94\u306e\u97f3\u6570\u3092\u5b88\u3063\u3066\u4f5c\u3063\u3066\u304f\u3060\u3055\u3044\u3002\u305d\u308c\u305e\u308c\u306b\u4e00\u8a00\u306e\u8aac\u660e\u3092\u4ed8\u3051\u3066\u304f\u3060\u3055\u3044\u3002<\/p>\n<\/blockquote>\n<p><em>(Write three haiku on the theme of spring, keeping the 5-7-5 syllable pattern, each with a one-line explanation.)<\/em><\/p>\n<pre><code class=\"language-text\">\u6625\u306f\u3001\u82bd\u5439\u304f\u7a7a\u306b\u3002\n\n\u6625\u306f\u3001\u98a8\u304c\u3053\u3082\u308b\u9803\u3002\n\n\u6625\u306f\u3001\u9759\u3051\u3055\u306e\u4e2d\u306e\u751f\u3002\n\n\u305d\u308c\u305e\u308c\u3001\u6625\u306e\u610f\u3092\u8fbc\u3081\u3066\u3044\u307e\u3059\u3002\n<\/code><\/pre>\n<p><!-- \/lmw:lab --><\/p>\n<p><!-- lmw:peers --><\/p>\n<h2>Recent Models in the Same Size Class<\/h2>\n<p><em>Models with <\/em><em>up to 4B<\/em><em> parameters that Local Model Watch covered recently, listed by code from our article log for comparison. VRAM tiers are this site&#8217;s estimates; licenses are as stated on the model cards.<\/em><\/p>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<p class=\"lmw-table-hint\" style=\"margin:0 0 4px;font-size:0.85em;opacity:0.7;\">\u2192 Scroll horizontally to see all columns<\/p>\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Model<\/th>\n<th>Parameters<\/th>\n<th>Smallest VRAM tier<\/th>\n<th>License<\/th>\n<th>Our article<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>LiquidAI\/LFM2.5-350M-Diffusion-Exp<\/td>\n<td>425M<\/td>\n<td>4GB<\/td>\n<td>lfm1.0<\/td>\n<td><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/10\/04\/liquidai-lfm2-5-350m-diffusion-exp\/\">LFM2.5-350M-Diffusion-Exp Text Generation Model: 4GB+ VRAM<\/a> (2026-10-03)<\/td>\n<\/tr>\n<tr>\n<td>togethercomputer\/Tev1-0.8B-experimental<\/td>\n<td>873M<\/td>\n<td>4GB<\/td>\n<td>\u2014<\/td>\n<td><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/26\/tev1-08b-experimental-2\/\">Tev1-0.8B-experimental Text Generation Model: 4GB+ VRAM<\/a> (2026-09-25)<\/td>\n<\/tr>\n<tr>\n<td>pfnet\/plamo-3-610m-fin-instruct<\/td>\n<td>890M<\/td>\n<td>4GB<\/td>\n<td>plamo-community-license<\/td>\n<td><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/24\/plamo-3-610m-fin-instruct-2\/\">plamo-3-610m-fin-instruct Text Generation Model: 4GB+ VRAM<\/a> (2026-09-24)<\/td>\n<\/tr>\n<tr>\n<td>harshatheg\/Qwen-2.5-1B-RLCD<\/td>\n<td>1.5B<\/td>\n<td>4GB<\/td>\n<td>apache-2.0<\/td>\n<td><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/17\/qwen-25-1b-rlcd-mlx-constrained-decoding\/\">Qwen-2.5-1B-RLCD Text Generation Model: 4GB+ VRAM<\/a> (2026-09-16)<\/td>\n<\/tr>\n<tr>\n<td>tencent\/Simple-Attention-Sparsification<\/td>\n<td>4.0B<\/td>\n<td>12GB<\/td>\n<td>\u2014<\/td>\n<td><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/14\/tencent-simple-attention-sparsification-qwen3\/\">Simple-Attention-Sparsification Text Generation Model: 12GB+ VRAM<\/a> (2026-09-14)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p><!-- \/lmw:peers --><\/p>\n<h2>How to Get It<\/h2>\n<p>It supports diverse backends and formats such as Transformers, vLLM, SGLang, llama.cpp, Ollama, LM Studio, MLX, ArcLight, vLLM Ascend, and LiteRT-LM. An example command for serving with SGLang is as follows:<\/p>\n<pre><code class=\"language-bash\">pip install &quot;sglang[srt]&gt;=0.5.16&quot;\npython -m sglang.launch_server --model-path openbmb\/MiniCPM5-2B --port 30000\n<\/code><\/pre>\n<p><!-- lmw:variants --><\/p>\n<h2>Quantized and Converted Variants<\/h2>\n<div class=\"lmw-table-scroll\" tabindex=\"0\" style=\"overflow-x:auto;-webkit-overflow-scrolling:touch;max-width:100%;\">\n<p class=\"lmw-table-hint\" style=\"margin:0 0 4px;font-size:0.85em;opacity:0.7;\">\u2192 Scroll horizontally to see all columns<\/p>\n<table style=\"width:max-content;min-width:100%;border-collapse:collapse;\">\n<thead>\n<tr>\n<th>Added<\/th>\n<th>Publisher<\/th>\n<th>Format<\/th>\n<th>Repository<\/th>\n<th>Smallest VRAM tier (build, est. memory)<\/th>\n<\/tr>\n<\/thead>\n<tbody>\n<tr>\n<td>2026-09-18 (10 days after publication)<\/td>\n<td>openbmb<\/td>\n<td>GGUF<\/td>\n<td><a href=\"https:\/\/huggingface.co\/openbmb\/MiniCPM5-2B-GGUF\">openbmb\/MiniCPM5-2B-GGUF<\/a><\/td>\n<td>Q8_0 3.0GB (fits in 4GB VRAM)<\/td>\n<\/tr>\n<tr>\n<td>2026-09-18 (10 days after publication)<\/td>\n<td>openbmb<\/td>\n<td>MLX<\/td>\n<td><a href=\"https:\/\/huggingface.co\/openbmb\/MiniCPM5-2B-MLX\">openbmb\/MiniCPM5-2B-MLX<\/a><\/td>\n<td>MLX 1.6GB (fits in 4GB VRAM)<\/td>\n<\/tr>\n<tr>\n<td>2026-09-18 (10 days after publication)<\/td>\n<td>bartowski<\/td>\n<td>GGUF<\/td>\n<td><a href=\"https:\/\/huggingface.co\/bartowski\/MiniCPM5-2B-GGUF\">bartowski\/MiniCPM5-2B-GGUF<\/a><\/td>\n<td>Q8_0 3.0GB (fits in 4GB VRAM)<\/td>\n<\/tr>\n<tr>\n<td>2026-09-18 (10 days after publication)<\/td>\n<td>openbmb<\/td>\n<td>GPTQ<\/td>\n<td><a href=\"https:\/\/huggingface.co\/openbmb\/MiniCPM5-2B-GPTQ\">openbmb\/MiniCPM5-2B-GPTQ<\/a><\/td>\n<td>GPTQ 2.3GB (fits in 4GB VRAM)<\/td>\n<\/tr>\n<tr>\n<td>2026-09-18 (10 days after publication)<\/td>\n<td>mlx-community<\/td>\n<td>MLX<\/td>\n<td><a href=\"https:\/\/huggingface.co\/mlx-community\/MiniCPM5-2B-8bit\">mlx-community\/MiniCPM5-2B-8bit<\/a><\/td>\n<td>MLX 8bit 3.0GB (fits in 4GB VRAM)<\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n<p>In addition, 43 converted build(s) from other uploaders exist on Hugging Face; this site lists only builds from the model&#8217;s publisher or established quantization maintainers.<\/p>\n<p><em>This section is appended automatically by Local Model Watch when a converted build of this model appears after publication. Memory figures are estimated from the size of the distributed files. What the quantization names mean: <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/glossary-quantization-en\/\">glossary<\/a>.<\/em><\/p>\n<p><!-- \/lmw:variants --><\/p>\n<p><!-- lmw:related --><\/p>\n<h2>Related Articles<\/h2>\n<ul>\n<li><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/10\/05\/qwen-3-8-flash-next-released\/\">Qwen3.8-Flash-Next Multimodal MoE Model: ~402GB Memory<\/a><\/li>\n<li><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/26\/deepseek-v4-pro-0813-released\/\">DeepSeek-V4-Pro-0813 Text Generation Model: ~998GB Memory, GGUF Builds<\/a><\/li>\n<li><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/23\/ifm-k2-horizon-32b-nvfp4-released\/\">K2-Horizon-32B-NVFP4: Our Test Answers, 32GB+ VRAM<\/a><\/li>\n<li><a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/2026\/09\/23\/ifm-k2-horizon-375b-a23b-nvfp4-released-2\/\">K2-Horizon-375B-A23B-NVFP4 Text Generation Model: ~257GB Memory<\/a><\/li>\n<\/ul>\n<p><!-- \/lmw:related --><\/p>\n<p><!-- lmw:next-steps --><\/p>\n<h2>What to Read Next<\/h2>\n<ul>\n<li><strong>Find models by VRAM<\/strong> (This model runs from the 4GB tier) \u2192 <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/vram-guide-8gb-en\/\">Other models that run on a 8GB GPU<\/a><\/li>\n<li><strong>Explore the same model family<\/strong> \u2192 <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/model-openbmb-minicpm5-en\/\">MiniCPM5 family overview (1 articles, 5 converted builds)<\/a><\/li>\n<li><strong>Engines that run this model<\/strong> \u2192 <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/engine-llama-cpp-en\/\">llama.cpp<\/a> \/ <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/engine-ollama-en\/\">Ollama<\/a> \/ <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/engine-vllm-en\/\">vLLM<\/a><\/li>\n<li><strong>What Q8_0 mean and where to get this model<\/strong> \u2192 <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/format-gguf-en\/\">GGUF format guide and models<\/a> \/ <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/format-mlx-en\/\">MLX format guide and models<\/a> \/ <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/format-gptq-en\/\">GPTQ format guide and models<\/a> \/ <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/glossary-quantization-en\/\">Quantization and model-format glossary<\/a><\/li>\n<li><strong>Learn about the publisher<\/strong> \u2192 <a href=\"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/publisher-openbmb-en\/\">OpenBMB: models, licenses and articles<\/a><\/li>\n<\/ul>\n<p><!-- \/lmw:next-steps --><\/p>\n<h2>Sources<\/h2>\n<ul>\n<li><a href=\"https:\/\/huggingface.co\/openbmb\/MiniCPM5-2B\">https:\/\/huggingface.co\/openbmb\/MiniCPM5-2B<\/a><\/li>\n<li><a href=\"https:\/\/arxiv.org\/pdf\/2602.09003\">https:\/\/arxiv.org\/pdf\/2602.09003<\/a><\/li>\n<li><a href=\"https:\/\/ultradata.openbmb.cn\/\">https:\/\/ultradata.openbmb.cn\/<\/a><\/li>\n<li><a href=\"https:\/\/huggingface.co\/datasets\/openbmb\/UltraX-Preview\">https:\/\/huggingface.co\/datasets\/openbmb\/UltraX-Preview<\/a><\/li>\n<li><a href=\"https:\/\/huggingface.co\/datasets\/openbmb\/UltraData-Code\">https:\/\/huggingface.co\/datasets\/openbmb\/UltraData-Code<\/a><\/li>\n<li><a href=\"https:\/\/huggingface.co\/datasets\/openbmb\/UltraData-SFT-Agent-2609\">https:\/\/huggingface.co\/datasets\/openbmb\/UltraData-SFT-Agent-2609<\/a><\/li>\n<li><a href=\"https:\/\/huggingface.co\/datasets\/openbmb\/UltraData-RL-2609\">https:\/\/huggingface.co\/datasets\/openbmb\/UltraData-RL-2609<\/a><\/li>\n<li><a href=\"https:\/\/huggingface.co\/datasets\/openbmb\/Ultra-FineWeb\">https:\/\/huggingface.co\/datasets\/openbmb\/Ultra-FineWeb<\/a><\/li>\n<li><a href=\"https:\/\/huggingface.co\/datasets\/openbmb\/Ultra-FineWeb-L3\">https:\/\/huggingface.co\/datasets\/openbmb\/Ultra-FineWeb-L3<\/a><\/li>\n<li><a href=\"https:\/\/huggingface.co\/datasets\/openbmb\/UltraData-Math\">https:\/\/huggingface.co\/datasets\/openbmb\/UltraData-Math<\/a><\/li>\n<li><a href=\"https:\/\/huggingface.co\/datasets\/openbmb\/UltraData-SFT-2605\">https:\/\/huggingface.co\/datasets\/openbmb\/UltraData-SFT-2605<\/a><\/li>\n<li><a href=\"https:\/\/flagos.io\/\">https:\/\/flagos.io\/<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/flagos-ai\/FlagGems\">https:\/\/github.com\/flagos-ai\/FlagGems<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/flagos-ai\/vllm-plugin-FL\">https:\/\/github.com\/flagos-ai\/vllm-plugin-FL<\/a><\/li>\n<li><a href=\"https:\/\/panhaoxuan.notion.site\/justrl-ii-scaling-small-llms-to-128k-reasoning-with-a-critic\">https:\/\/panhaoxuan.notion.site\/justrl-ii-scaling-small-llms-to-128k-reasoning-with-a-critic<\/a><\/li>\n<li><a href=\"https:\/\/huggingface.co\/openbmb\/MiniCPM5-1B\">https:\/\/huggingface.co\/openbmb\/MiniCPM5-1B<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B\">https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B-SFT\">https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B-SFT<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B-Midtrain\">https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B-Midtrain<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B-Base\">https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B-Base<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B-GGUF\">https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B-GGUF<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B-MLX\">https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B-MLX<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B-GPTQ\">https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B-GPTQ<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B-DSpark\">https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B-DSpark<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B-DSpark-GGUF\">https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-2B-DSpark-GGUF<\/a><\/li>\n<li><a href=\"https:\/\/huggingface.co\/litert-community\/MiniCPM5-2B\">https:\/\/huggingface.co\/litert-community\/MiniCPM5-2B<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-1B\">https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-1B<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-1B-SFT\">https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-1B-SFT<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-1B-Base\">https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-1B-Base<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-1B-GGUF\">https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-1B-GGUF<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-1B-MLX\">https:\/\/www.modelscope.cn\/models\/OpenBMB\/MiniCPM5-1B-MLX<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/transformers.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/transformers.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-transformers\/SKILL.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-transformers\/SKILL.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/vllm.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/vllm.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-vllm\/SKILL.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-vllm\/SKILL.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/sglang.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/sglang.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-sglang\/SKILL.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-sglang\/SKILL.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/llama_cpp.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/llama_cpp.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-llama-cpp\/SKILL.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-llama-cpp\/SKILL.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/ollama.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/ollama.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-ollama\/SKILL.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-ollama\/SKILL.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/lmstudio.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/lmstudio.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-lmstudio\/SKILL.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-lmstudio\/SKILL.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/mlx.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/mlx.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-mlx\/SKILL.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-mlx\/SKILL.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/arclight.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/arclight.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-arclight\/SKILL.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-arclight\/SKILL.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/vllm_ascend.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/vllm_ascend.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-vllm-ascend\/SKILL.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-vllm-ascend\/SKILL.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/litert.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/deployment\/litert.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-litert\/SKILL.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-deploy-litert\/SKILL.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/finetune\/trl.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/finetune\/trl.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-finetune-trl\/SKILL.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-finetune-trl\/SKILL.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/finetune\/llamafactory.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/finetune\/llamafactory.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-finetune-llamafactory\/SKILL.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-finetune-llamafactory\/SKILL.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/finetune\/ms_swift.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/finetune\/ms_swift.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-finetune-ms-swift\/SKILL.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-finetune-ms-swift\/SKILL.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/finetune\/unsloth.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/docs\/finetune\/unsloth.md<\/a><\/li>\n<li><a href=\"https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-finetune-unsloth\/SKILL.md\">https:\/\/github.com\/OpenBMB\/MiniCPM\/blob\/main\/skills\/minicpm5-finetune-unsloth\/SKILL.md<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-nvidia-FlagOS\">https:\/\/www.modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-nvidia-FlagOS<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-hygon-FlagOS\">https:\/\/www.modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-hygon-FlagOS<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-metax-FlagOS\">https:\/\/www.modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-metax-FlagOS<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-iluvatar-FlagOS\">https:\/\/www.modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-iluvatar-FlagOS<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-zhenwu-FlagOS\">https:\/\/www.modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-zhenwu-FlagOS<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-mthreads-FlagOS\">https:\/\/www.modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-mthreads-FlagOS<\/a><\/li>\n<li><a href=\"https:\/\/www.modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-kunlunxin-FlagOS\">https:\/\/www.modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-kunlunxin-FlagOS<\/a><\/li>\n<li><a href=\"https:\/\/modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-ascend-FlagOS\">https:\/\/modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-ascend-FlagOS<\/a><\/li>\n<li><a href=\"https:\/\/modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-Armv9-FlagOS\">https:\/\/modelscope.cn\/models\/FlagRelease\/MiniCPM5-2B-Armv9-FlagOS<\/a><\/li>\n<\/ul>\n<p><!-- lmw:updates --><\/p>\n<h2>Update History<\/h2>\n<ul>\n<li>2026-09-18: Added converted builds to \u201cQuantized and Converted Variants\u201d: openbmb\/MiniCPM5-2B-GGUF, openbmb\/MiniCPM5-2B-MLX, bartowski\/MiniCPM5-2B-GGUF, openbmb\/MiniCPM5-2B-GPTQ, mlx-community\/MiniCPM5-2B-8bit<\/li>\n<li>2026-09-18: Updated the hardware requirements table with the actual file sizes of openbmb\/MiniCPM5-2B-GGUF.<\/li>\n<li>2026-09-24: Rebuilt the article: Added a &#8220;can you run it locally?&#8221; verdict from the distribution formats, converted builds, engine registries, license terms and measured compression<\/li>\n<li>2026-09-25: Changed the title to show what the article covers (VRAM requirements, file list, etc.).<\/li>\n<li>2026-09-28: Added our own measurements: Japanese token efficiency, what is inside the GGUF file.<\/li>\n<li>2026-09-28: Added our own measurements: CPU speed and answers to Japanese questions.<\/li>\n<li>2026-09-29: Added our own measurements: quality loss by quantization level.<\/li>\n<li>2026-09-30: Changed the title to show what the article covers (VRAM requirements, file list, etc.).<\/li>\n<\/ul>\n<p><!-- \/lmw:updates --><\/p>\n","protected":false},"excerpt":{"rendered":"<p>OpenBMB has released MiniCPM5-2B, a 2B-class on-device model delivering SOTA performance in coding, math, long context, and agent tasks.<\/p>\n","protected":false},"author":1,"featured_media":7639,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[310],"tags":[163,136,318,320,169,592,117],"class_list":["post-155","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-new-models","tag-gguf-en","tag-llm-en","tag-minicpm5-2b-en","tag-openbmb-en","tag-sglang-en","tag-vllm-en","tag--en"],"lang":"en","translations":{"en":155,"ja":154},"_links":{"self":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts\/155","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/comments?post=155"}],"version-history":[{"count":83,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts\/155\/revisions"}],"predecessor-version":[{"id":14813,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/posts\/155\/revisions\/14813"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/media\/7639"}],"wp:attachment":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/media?parent=155"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/categories?post=155"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/tags?post=155"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}