{"id":1202,"date":"2026-09-17T06:20:39","date_gmt":"2026-09-16T21:20:39","guid":{"rendered":"https:\/\/localmodelwatch.tsuchitsuchi.com\/vram-guide-en\/"},"modified":"2026-11-02T23:08:04","modified_gmt":"2026-11-02T14:08:04","slug":"vram-guide-en","status":"publish","type":"page","link":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/vram-guide-en\/","title":{"rendered":"Local Models by VRAM: Quick Reference"},"content":{"rendered":"

Open-weight models covered by Local Model Watch, grouped by the smallest amount of GPU memory (VRAM) they are estimated to run in. Each figure is computed by this site from the actual size of the distributed files plus a runtime overhead \u2014 not quoted from model cards. See our editorial policy<\/a> for the method.<\/p>\n

Models listed under a tier do not<\/strong> fit in the smaller tiers. If you have a 24GB GPU, the 8GB, 12GB, 16GB and 24GB tables all apply to you. Models that don’t fit even the largest tier (80GB) are listed separately at the bottom, since that too is something readers need to know.<\/p>\n

The Runs on<\/strong> column lists engines (Ollama \/ LM Studio via llama.cpp, vLLM, MLX) confirmed to run the model today, matched from each engine’s own model registry and the GGUF\/MLX builds we have found; it does not apply to image, video and audio models. “-” means either not applicable or not yet confirmed \u2014 see the model’s own article for the reasoning.<\/p>\n

Last updated 2026-11-02 (JST). 52 models listed.<\/em><\/p>\n

Everything that fits your GPU: 8GB<\/a> \/ 12GB<\/a> \/ 16GB<\/a> \/ 24GB<\/a><\/p>\n

4GB (laptop iGPU \/ phone class)<\/h2>\n
\n

\u2192 Scroll horizontally to see all columns<\/p>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
Model<\/th>\nParameters<\/th>\nSmallest build<\/th>\nEst. memory<\/th>\nRuns on<\/th>\nArticle<\/th>\n<\/tr>\n<\/thead>\n
LiquidAI\/LFM2.5-350M-Diffusion-Exp<\/td>\n425M<\/td>\nBF16 (0.8GB)<\/td>\n0.9GB<\/td>\n–<\/td>\nRead<\/a><\/td>\n<\/tr>\n
Cloudflare\/clef-flash<\/td>\n9.4B<\/td>\nIQ2_M (3.3GB)<\/td>\n4.0GB<\/td>\nOllama \/ LM Studio, vLLM, MLX<\/td>\nRead<\/a><\/td>\n<\/tr>\n
FermionResearch\/Phonon-2<\/td>\n627M<\/td>\nQ8_0 (0.7GB)<\/td>\n0.8GB<\/td>\n–<\/td>\nRead<\/a><\/td>\n<\/tr>\n
apple\/LensVLM-9B<\/td>\n9.4B<\/td>\nIQ2_M (3.3GB)<\/td>\n4.0GB<\/td>\nOllama \/ LM Studio, vLLM, MLX<\/td>\nRead<\/a><\/td>\n<\/tr>\n
togethercomputer\/Tev1-0.8B-experimental<\/td>\n873M<\/td>\nF32 (1.6GB)<\/td>\n2.0GB<\/td>\nvLLM<\/td>\nRead<\/a><\/td>\n<\/tr>\n
LiquidAI\/LFM2.5-VL-3B-DSpark<\/td>\n279M<\/td>\nF16 (0.5GB)<\/td>\n0.6GB<\/td>\n–<\/td>\nRead<\/a><\/td>\n<\/tr>\n
togethercomputer\/Tev1-4B-experimental<\/td>\n4.7B<\/td>\nQ6_K_S (3.4GB)<\/td>\n4.0GB<\/td>\nOllama \/ LM Studio, vLLM<\/td>\nRead<\/a><\/td>\n<\/tr>\n
pfnet\/plamo-3-610m-fin-instruct<\/td>\n890M<\/td>\nF16 (1.7GB)<\/td>\n2.0GB<\/td>\nOllama \/ LM Studio, vLLM<\/td>\nRead<\/a><\/td>\n<\/tr>\n
Cactus-Compute\/needle3<\/td>\n–<\/td>\nOriginal precision (0.2GB)<\/td>\n0.3GB<\/td>\n–<\/td>\nRead<\/a><\/td>\n<\/tr>\n
tencent\/WeVisDoc-4B<\/td>\n4.4B<\/td>\nQ5_K_S (2.9GB)<\/td>\n3.5GB<\/td>\n–<\/td>\nRead<\/a><\/td>\n<\/tr>\n
harshatheg\/Qwen-2.5-1B-RLCD<\/td>\n1.5B<\/td>\nBF16 (2.9GB)<\/td>\n3.5GB<\/td>\nMLX<\/td>\nRead<\/a><\/td>\n<\/tr>\n
Efficient-Large-Model\/H3-to-LTX-Latent-Adapter<\/td>\n195M<\/td>\nBF16 (0.4GB)<\/td>\n0.4GB<\/td>\n–<\/td>\nRead<\/a><\/td>\n<\/tr>\n
openbmb\/MiniCPM5-2B<\/td>\n2.5B<\/td>\nQ8_0 (2.5GB)<\/td>\n3.0GB<\/td>\nOllama \/ LM Studio, vLLM, MLX<\/td>\nRead<\/a><\/td>\n<\/tr>\n
phasefield-audio\/Irodori-TTS-v4.1-Anime<\/td>\n766M<\/td>\nF32 (2.9GB)<\/td>\n3.4GB<\/td>\n–<\/td>\nRead<\/a><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

8GB (RTX 4060 \/ 3060 Ti, etc.)<\/h2>\n
\n

\u2192 Scroll horizontally to see all columns<\/p>\n\n\n\n\n\n\n\n
Model<\/th>\nParameters<\/th>\nSmallest build<\/th>\nEst. memory<\/th>\nRuns on<\/th>\nArticle<\/th>\n<\/tr>\n<\/thead>\n
google\/DiarizationLM-Gemma-4-E4B-v1<\/td>\n8.0B<\/td>\nQ4_K_M (4.9GB)<\/td>\n5.9GB<\/td>\nOllama \/ LM Studio, vLLM<\/td>\nRead<\/a><\/td>\n<\/tr>\n
Qwen\/Qwen3.8-27B<\/td>\n27.8B<\/td>\nIQ1_M (6.3GB)<\/td>\n7.5GB<\/td>\nOllama \/ LM Studio, vLLM, MLX<\/td>\nRead<\/a><\/td>\n<\/tr>\n
prism-ml\/Ternary-Bonsai-2-27B-gguf<\/td>\n27.8B<\/td>\nQ1_0 (5.5GB)<\/td>\n6.6GB<\/td>\nOllama \/ LM Studio, MLX<\/td>\nRead<\/a><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

12GB (RTX 4070 \/ 3060 12GB, etc.)<\/h2>\n
\n

\u2192 Scroll horizontally to see all columns<\/p>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
Model<\/th>\nParameters<\/th>\nSmallest build<\/th>\nEst. memory<\/th>\nRuns on<\/th>\nArticle<\/th>\n<\/tr>\n<\/thead>\n
Cloudflare\/clef<\/td>\n27.4B<\/td>\nQ2_K (9.9GB)<\/td>\n11.8GB<\/td>\nOllama \/ LM Studio, vLLM, MLX<\/td>\nRead<\/a><\/td>\n<\/tr>\n
Altworld\/Hemmingway-1<\/td>\n26.9B<\/td>\nIQ2_M (9.8GB)<\/td>\n11.8GB<\/td>\nOllama \/ LM Studio, vLLM, MLX<\/td>\nRead<\/a><\/td>\n<\/tr>\n
Edge0\/Audio8-ASR-Infinite<\/td>\n4.1B<\/td>\nBF16 (7.6GB)<\/td>\n9.1GB<\/td>\n–<\/td>\nRead<\/a><\/td>\n<\/tr>\n
ggml-org\/MiMo-V2.6-Distill-Qwen-9B-GGUF<\/td>\n9.4B<\/td>\nQ8_0 (8.9GB)<\/td>\n10.6GB<\/td>\nOllama \/ LM Studio<\/td>\nRead<\/a><\/td>\n<\/tr>\n
bartowski\/vectionlabs_Salience-27B-R6-GGUF<\/td>\n27.8B<\/td>\nIQ2_M (9.8GB)<\/td>\n11.8GB<\/td>\nOllama \/ LM Studio<\/td>\nRead<\/a><\/td>\n<\/tr>\n
tencent\/Simple-Attention-Sparsification<\/td>\n4.0B<\/td>\nBF16 (7.5GB)<\/td>\n9.0GB<\/td>\n–<\/td>\nRead<\/a><\/td>\n<\/tr>\n
bartowski\/TheDrummer_Orion-26B-A4B-v1.1-GGUF<\/td>\n25.8B<\/td>\nIQ2_S (9.6GB)<\/td>\n11.5GB<\/td>\nOllama \/ LM Studio<\/td>\nRead<\/a><\/td>\n<\/tr>\n
Comfy-Org\/YuE2<\/td>\n3.6B<\/td>\nBF16 (6.8GB)<\/td>\n8.1GB<\/td>\n–<\/td>\nRead<\/a><\/td>\n<\/tr>\n
bartowski\/Gryphe_Pantheon-Reasoning-26B-A4B-1.1-V2-GGUF<\/td>\n26.5B<\/td>\nIQ2_S (10.0GB)<\/td>\n12.0GB<\/td>\nOllama \/ LM Studio<\/td>\nRead<\/a><\/td>\n<\/tr>\n
m-a-p\/YuE2-3B<\/td>\n3.6B<\/td>\nBF16 (6.8GB)<\/td>\n8.1GB<\/td>\n–<\/td>\nRead<\/a><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

16GB (RTX 5060 Ti 16GB \/ 4060 Ti 16GB, etc.)<\/h2>\n
\n

\u2192 Scroll horizontally to see all columns<\/p>\n\n\n\n\n\n\n\n\n
Model<\/th>\nParameters<\/th>\nSmallest build<\/th>\nEst. memory<\/th>\nRuns on<\/th>\nArticle<\/th>\n<\/tr>\n<\/thead>\n
orcarouter\/OrcaSAQ-2-27B<\/td>\n27.8B<\/td>\nEXL3 (11.4GB)<\/td>\n13.7GB<\/td>\nvLLM<\/td>\nRead<\/a><\/td>\n<\/tr>\n
agentionai\/Signal-3.8-27B-GGUF<\/td>\n27.8B<\/td>\nIQ4_XS (13.3GB)<\/td>\n15.9GB<\/td>\nOllama \/ LM Studio<\/td>\nRead<\/a><\/td>\n<\/tr>\n
nex-agi\/Nex-N2.5-mini<\/td>\n35.1B<\/td>\nQ2_K (12.9GB)<\/td>\n15.5GB<\/td>\nOllama \/ LM Studio, vLLM, MLX<\/td>\nRead<\/a><\/td>\n<\/tr>\n
Jackrong\/Qwopus3.8-27B-Flash-GGUF<\/td>\n27.8B<\/td>\nQ3_K_M (12.6GB)<\/td>\n15.1GB<\/td>\nOllama \/ LM Studio<\/td>\nRead<\/a><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

24GB (RTX 4090 \/ 3090, etc.)<\/h2>\n
\n

\u2192 Scroll horizontally to see all columns<\/p>\n\n\n\n\n\n\n\n\n\n\n
Model<\/th>\nParameters<\/th>\nSmallest build<\/th>\nEst. memory<\/th>\nRuns on<\/th>\nArticle<\/th>\n<\/tr>\n<\/thead>\n
nvidia\/PixelUMM<\/td>\n8.2B<\/td>\nBF16 (15.3GB)<\/td>\n18.3GB<\/td>\n–<\/td>\nRead<\/a><\/td>\n<\/tr>\n
ggml-org\/OpenJev-GGUF<\/td>\n27.4B<\/td>\nQ4_K_M (17.7GB)<\/td>\n21.2GB<\/td>\nOllama \/ LM Studio<\/td>\nRead<\/a><\/td>\n<\/tr>\n
XingChen-AGI\/Xing4.0-29B-A4B<\/td>\n31.2B<\/td>\nIQ4_NL (18.7GB)<\/td>\n22.5GB<\/td>\nOllama \/ LM Studio, MLX<\/td>\nRead<\/a><\/td>\n<\/tr>\n
inclusionAI\/Realtime-Venus<\/td>\n–<\/td>\nOriginal precision(Realtime-Venus-Audio) (17.5GB)<\/td>\n20.9GB<\/td>\n–<\/td>\nRead<\/a><\/td>\n<\/tr>\n
EleutherAI\/olmo3-7b-sdf-sft-clean150<\/td>\n7.3B<\/td>\nBF16 (13.6GB)<\/td>\n16.3GB<\/td>\nvLLM<\/td>\nRead<\/a><\/td>\n<\/tr>\n
Edge0\/Edge0-35B-A3B-preview<\/td>\n36.0B<\/td>\nMLX 4bit (18.2GB)<\/td>\n21.8GB<\/td>\nvLLM, MLX<\/td>\nRead<\/a><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

32GB (RTX 5090, etc.)<\/h2>\n
\n

\u2192 Scroll horizontally to see all columns<\/p>\n\n\n\n\n\n\n\n
Model<\/th>\nParameters<\/th>\nSmallest build<\/th>\nEst. memory<\/th>\nRuns on<\/th>\nArticle<\/th>\n<\/tr>\n<\/thead>\n
MiniMaxAI\/MiniMax-H3<\/td>\n–<\/td>\nQ3_K_XL (26.5GB)<\/td>\n31.8GB<\/td>\n–<\/td>\nRead<\/a><\/td>\n<\/tr>\n
IFM\/K2-Horizon-32B-NVFP4<\/td>\n–<\/td>\nNVFP4 (21.7GB)<\/td>\n26.0GB<\/td>\nvLLM<\/td>\nRead<\/a><\/td>\n<\/tr>\n
IFM\/K2-Horizon-MoVA-36B-A4B-GGUF<\/td>\n37.4B<\/td>\nQ5_K_M (24.6GB)<\/td>\n29.5GB<\/td>\nOllama \/ LM Studio<\/td>\nRead<\/a><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

80GB class (A100 \/ H100)<\/h2>\n
\n

\u2192 Scroll horizontally to see all columns<\/p>\n\n\n\n\n\n
Model<\/th>\nParameters<\/th>\nSmallest build<\/th>\nEst. memory<\/th>\nRuns on<\/th>\nArticle<\/th>\n<\/tr>\n<\/thead>\n
elyza\/ELYZA-Thinking-1.0-llm-jp-4-32b-a3b<\/td>\n32.1B<\/td>\nBF16 (59.9GB)<\/td>\n71.8GB<\/td>\nvLLM<\/td>\nRead<\/a><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n

Over 80GB (datacenter GPU \/ multi-GPU required)<\/h2>\n
\n

\u2192 Scroll horizontally to see all columns<\/p>\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n\n
Model<\/th>\nParameters<\/th>\nSmallest build<\/th>\nEst. memory<\/th>\nRuns on<\/th>\nArticle<\/th>\n<\/tr>\n<\/thead>\n
Qwen\/Qwen3.8-Flash-Next<\/td>\n180.0B<\/td>\nBF16 (335.3GB)<\/td>\n402.3GB<\/td>\nvLLM<\/td>\nRead<\/a><\/td>\n<\/tr>\n
NaiveAI\/Naive-N0.5-Flash<\/td>\n–<\/td>\nOriginal precision (575.4GB)<\/td>\n690.5GB<\/td>\n–<\/td>\nRead<\/a><\/td>\n<\/tr>\n
ggml-org\/GLM-5.3-Flash-GGUF<\/td>\n321.3B<\/td>\nQ2_K (124.6GB)<\/td>\n149.5GB<\/td>\nOllama \/ LM Studio<\/td>\nRead<\/a><\/td>\n<\/tr>\n
XiaomiMiMo\/MiMo-V2.6-Pro-RL<\/td>\n–<\/td>\nQ2_K (347.8GB)<\/td>\n417.4GB<\/td>\nOllama \/ LM Studio, vLLM<\/td>\nRead<\/a><\/td>\n<\/tr>\n
deepseek-ai\/DeepSeek-V4-Pro-0813<\/td>\n1650.5B<\/td>\nFP8 (831.4GB)<\/td>\n997.7GB<\/td>\nOllama \/ LM Studio, vLLM<\/td>\nRead<\/a><\/td>\n<\/tr>\n
IFM\/K2-Horizon-375B-A23B-NVFP4<\/td>\n–<\/td>\nNVFP4 (213.8GB)<\/td>\n256.6GB<\/td>\nvLLM<\/td>\nRead<\/a><\/td>\n<\/tr>\n
ggml-org\/MiMo-V2.6-Flash-RL-GGUF<\/td>\n–<\/td>\nQ2_K (117.5GB)<\/td>\n141.1GB<\/td>\nOllama \/ LM Studio<\/td>\nRead<\/a><\/td>\n<\/tr>\n
bartowski\/Intern-S2-397B-GGUF<\/td>\n403.4B<\/td>\nIQ1_S (85.3GB)<\/td>\n102.4GB<\/td>\nOllama \/ LM Studio<\/td>\nRead<\/a><\/td>\n<\/tr>\n
deepseek-ai\/DeepSeek-V4.1-Flash<\/td>\n763.2B<\/td>\nFP8 (475.3GB)<\/td>\n570.3GB<\/td>\nvLLM<\/td>\nRead<\/a><\/td>\n<\/tr>\n
nvidia\/DeepSeek-V4-Pro-0813-nvfp4-DSpark<\/td>\n1650.5B<\/td>\nNVFP4 (837.4GB)<\/td>\n1004.9GB<\/td>\nvLLM<\/td>\nRead<\/a><\/td>\n<\/tr>\n
nex-agi\/Nex-N2.5-Pro<\/td>\n396.8B<\/td>\nIQ1_S (81.8GB)<\/td>\n98.2GB<\/td>\nOllama \/ LM Studio, vLLM<\/td>\nRead<\/a><\/td>\n<\/tr>\n<\/tbody>\n<\/table>\n<\/div>\n","protected":false},"excerpt":{"rendered":"

Open-weight models covered by Local Model Watch, grouped by the smallest amount of GPU memory (VRAM) they are estimated to run in. Each figure is computed by this site from the actual size of the distributed files plus a runtime overhead \u2014 not quoted from model cards. See our editorial policy for the method. Models […]<\/p>\n","protected":false},"author":1,"featured_media":0,"parent":0,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"footnotes":""},"class_list":["post-1202","page","type-page","status-publish","hentry"],"lang":"en","translations":{"en":1202,"ja":1200},"_links":{"self":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/pages\/1202","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/types\/page"}],"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=1202"}],"version-history":[{"count":95,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/pages\/1202\/revisions"}],"predecessor-version":[{"id":14750,"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/pages\/1202\/revisions\/14750"}],"wp:attachment":[{"href":"https:\/\/localmodelwatch.tsuchitsuchi.com\/en\/wp-json\/wp\/v2\/media?parent=1202"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}