MiMo-V2.6-Distill-Qwen-9B-GGUF: Our Test Answers, 12GB+ VRAM

September 22, 2026November 9, 2026

At a Glance

Values determined by this site’s code when the information was collected. Dates are JST.

What we checked ourselves

• Ran MiMo-V2.6-Distill-Qwen-9B-Q8_0.gguf (Q8_0) on our CPU-only server (Neoverse-N1, 3 threads, llama.cpp b11223): generation at 3.7 tokens/s (a CPU figure, not a GPU speed).
• Its answers to our fixed Japanese questions are shown verbatim below.
• It needs about 21% fewer tokens than the Qwen3 tokenizer for the same Japanese text, so about 1.26× as much Japanese fits in the same context length, and generation is faster per character.

Details and conditions are in “Our Own Measurements” below.

Overview

On September 21, 2026, ggml-org released the GGUF version of “MiMo-V2.6-Distill-Qwen-9B", an agent-oriented model developed by Xiaomi MiMo. Based on Qwen3.5-9B, this 9.4B parameter model has undergone distillation and supervised fine-tuning (SFT) using datasets generated by MiMo. It is designed to target four main domains: coding, general agent tasks, visual coding, and cybersecurity.

According to the model card, it was released as an open research starting point for agentic reinforcement learning. The GGUF distribution also includes a Q8_0 mmproj file for the vision encoder, providing support for multimodal tasks. Readers can try out this agent-specialized model in GGUF-compatible environments such as llama.cpp and Ollama.

Specifications

• Parameters: 9.4B
• Architecture: Qwen3_5ForConditionalGeneration (Qwen 3.5)

Performance

The performance evaluations shown below are figures based on the technical report of the original model (XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B) released by Xiaomi MiMo, featuring comparisons with the base model Qwen3.5-9B. Note that in the GGUF quantization format, these figures may vary slightly due to the quantization process.

† Figures from internal evaluation sets

From the publisher’s measurement results, it can be seen that this model improves scores across almost all evaluation items compared to the base model Qwen3.5-9B. Notably, the score increase on “Terminal Bench 2.1" from 27.0 to 37.1 is worth highlighting. Terminal-Bench is an indicator that measures practical ability as an agent—whether it can actually execute commands on a terminal and see a given task through to completion—suggesting that this model is strong in tasks involving terminal operations.

In addition, dramatic improvements are seen in specific specialized areas, such as moving from 32.0 to 44.6 on SWE Pro, a challenging benchmark in the coding field, and from 5.7 to 31.3 on the internal cyber security evaluation (MiMo Cyber mini). On the other hand, items like SWE Verified showed only a marginal improvement from 60.0 to 61.1, indicating that it does not show a uniformly overwhelming difference across all coding tasks. Overall, it can be said to be a model that demonstrates particular strength in areas requiring agentic behavior, such as automation (AutomationBench) and tool use (Toolathlon-Verified).

For reference, the composition of the data used to train this model is reported as follows. Out of a total 77.4B tokens, loss-bearing tokens contributing to learning account for 27.2B tokens.

Strengths and Use Cases

The greatest identity of this model lies in its advanced autonomy as an “agent" that goes beyond mere text generation. As developer Xiaomi MiMo positions this model as a research foundation for agentic reinforcement learning, it excels in the ability to interpret complex instructions, break them down into concrete steps, and execute them. By applying supervised fine-tuning (SFT) using high-quality datasets generated by MiMo to the base model Qwen3.5-9B, it has achieved dramatic evolution in specific specialized domains.

Particularly noteworthy is its adaptability to tasks involving terminal operations. The high score on “Terminal-Bench 2.1" mentioned in the performance section indicates that this model can accurately simulate or direct command-line operations in local or server environments. It holds the potential to directly support engineers’ daily workflows, such as creating system administration automation scripts, building complex deployment procedures, and even interactive system troubleshooting. Regarding tool-use, as indicated by the score improvement in Toolathlon-Verified, its capability to properly invoke external tools to complete tasks has been enhanced.

Furthermore, the model is designed to explicitly output a “thinking" process. This means the model takes internal reasoning steps before producing a final answer, which has the effect of preventing logical breakdowns, especially in mathematical calculations and complex algorithm construction. Users can enable the thinking function via API to check reasoning_content and see what logical progression the model took to reach a conclusion. This transparency is extremely useful in debugging tasks or critical scenes where humans need to verify the grounds for the model’s decisions.

The breadth of supported domains and specialization is also noteworthy. The training data strategically blends the following four areas:

• Fusion of Coding and Vision: Through the enhancement of the “Visual Coding" domain, it supports development tasks starting from visual information, such as reading website screenshots or UI blueprints and generating HTML/CSS or React code to realize them. 27.4% of the training data is allocated to visual-related content (Visual), expecting advanced coordination between image recognition and code generation.
• Cybersecurity: Cybersecurity-related data amounting to 11 billion tokens (11.0B tokens) has been fed into the model. This anticipates applications in highly specialized security tasks that are difficult for general LLMs, such as assisting with vulnerability diagnostics, analyzing security logs, and creating penetration testing scenarios.
• General Agent Tasks: As shown by score improvements in benchmarks like OfficeQA and JobBench, practical agent functions such as office automation and job process management have been enhanced.

Hardware Requirements

Estimated requirements (calculated by Local Model Watch) — 9.4B parameters (taken from the base model XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B)

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’s authors. Compare with other models in our VRAM quick reference. What the quantization names mean: glossary.

Can You Run It Locally?

Runs in Ollama, LM Studio and llama.cpp as-is.

It is distributed in GGUF, so no conversion is needed.

License — mit (Commercial use allowed): Permits commercial use, modification and redistribution, provided the copyright notice and license text are retained.

Compression: the Q8_0 build measures 8.12 bits per weight — about 51% the size of the original 16-bit weights, calculated by this site from the actual file sizes.

Compiled by this site’s code from the published formats, converted builds we have found, and each engine’s own model registry. “Not found" means we have not seen such a build, not that none exists. License summaries are not legal advice — check the publisher’s original terms before relying on them.

Our Own Measurements

Values we measured ourselves on our server (no GPU) by actually reading and running this model’s files — not figures copied from the model card. How we measure, and the results for every model: Our Measurements.

What We Found

When running the Q8_0 file of MiMo-V2.6-Distill-Qwen-9B-GGUF on a Neoverse-N1 CPU with 3 threads, the generation speed was 3.69 tokens per second, which translates to 6.8 Japanese characters per second. This speed is close to Cloudflare_clef-flash at 3.85 tokens per second measured under the same conditions, and it consumed 14.8GB of memory at peak. Additionally, the number of tokens per 1,000 Japanese characters is 546, allowing it to process Japanese more efficiently with fewer tokens compared to 688 for Qwen3.

For the five Japanese questions, answers were obtained for all of them without hitting the upper limit truncation, repeating the same lines, or mixing in thinking processes. In the summarization of an article of about 1,200 characters, it outputted the exact three sentences as instructed, and in the shopping word problem, it correctly answered 280 yen. Furthermore, in image reading, it correctly answered the busiest day of the week, the 260 visitors on Saturday, and the weekday total of 635 visitors in the graph of visitors by day of the week.

On the other hand, in the knowledge question about listing prefectures not facing the ocean, while it listed 3 out of the 8 prefectures not facing the ocean, it mistakenly included 11 prefectures that actually do face the ocean. In drafting a polite email to a business partner, an answer was successfully generated, but two English words were mixed into the Japanese response. Overall, while calculations and data reading were handled accurately, challenges remained in accurately grasping geographical knowledge.

Japanese Token Efficiency

It needs about 21% fewer tokens than the Qwen3 tokenizer for the same Japanese text, so about 1.26× as much Japanese fits in the same context length, and generation is faster per character.

Counted with the tokenizer.json of XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B 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.

Inside the GGUF File

File: MiMo-V2.6-Distill-Qwen-9B-Q8_0.gguf (8.87GB, Q8_0). We read only the header (metadata) of the file, not the weights.

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.

Running It on a CPU Only

Conditions: Neoverse-N1, 3 threads, no GPU, llama.cpp b11223, Q8_0 (8.87GB) from ggml-org/MiMo-V2.6-Distill-Qwen-9B-GGUF, measured 2026-09-30 (JST). This repository offers no 4-bit file, so we measured the Q8_0 file it provides. Other models measured under the same conditions (closest in file size):

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.

Answers to Our Japanese Questions (Verbatim)

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.

Summarization

次の記事を3文で要約してください。

みどり市は今年の春、市内に三つある公立図書館のうち最も古い中央図書館の改修を終え、十月一日に再開館した。改修は約一年半にわたり、総事業費は約十二億円。このうち約四億円は国の補助金でまかなわれた。

最大の変化は、閲覧席の配置である。以前は館内の座席の大半が学習用の机で、試験期間になると中高生で埋まり、高齢の利用者や小さな子ども連れが座れないという苦情が毎年寄せられていた。改修後は座席数を百二十席から二百席に増やしたうえで、学習用の机を三階にまとめ、一階は会話ができる「にぎわいフロア」、二階は静かに本を読む「閲覧フロア」に分けた。一階には予約制のグループ学習室が四室と、飲み物を持ち込めるカフェスペースも設けられた。

蔵書の面でも見直しが進んだ。貸出の少ない古い資料約三万冊を市内の他の館や保存書庫に移し、代わりに子ども向けの絵本と、介護・子育て・仕事探しといった暮らしに関わる実用書を約一万五千冊増やした。電子書籍の貸出サービスも始まり、利用登録をすれば自宅から一度に三冊まで、二週間借りられる。

建物そのものにも手が入った。築四十年を超えていた空調設備を入れ替え、屋上には太陽光パネルを設置して、館内で使う電力の約二割をまかなえるようにした。車いすの利用者から要望の多かったエレベーターは一基から二基に増え、書架の間の通路も以前より三十センチ広げられた。入口には段差のないスロープが新たに設けられ、授乳室と多目的トイレも各階に一つずつ置かれた。

一方で課題も残る。駐車場は改修前と同じ四十台のままで、週末には近くの道路に順番待ちの車が並んだ。市は隣接する市有地を臨時駐車場として開放したが、恒久的な対策は決まっていない。また、開館時間を平日の夜九時まで延ばしたことで職員の負担が増えており、職員団体は人員の追加を求めている。

利用者の受け止めはおおむね好意的だ。週に二度、孫を連れて来るという七十代の女性は「前は子どもが声を出すと周りの目が気になったが、一階なら気兼ねなく絵本を読んであげられる」と話す。一方、受験を控えた高校生の中には「学習席が三階だけになり、夕方はすぐに満席になる」と不満を漏らす人もいる。図書館は、学習席の空き状況をホームページで確認できる仕組みを年内に導入する予定だ。

再開館に合わせて、催しも増えた。毎週土曜の午前には読み聞かせ会、月に一度は地元の農家や商店主を講師に招く「まちの先生」講座が開かれる。十一月には、市内の中学生が選んだ本を紹介する展示も予定されている。

館長の山本和子さんは「本を借りる場所から、人が集まって過ごす場所へ変えたかった。再開館からの一か月で来館者は以前の一・八倍になった」と話す。市は来年度、残る二つの図書館についても利用者への聞き取りを行い、改修の必要性を検討する方針だ。

(Summarize the following article in three sentences. (The article, about 1,200 Japanese characters, describes the renovation and reopening of a city’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’ reactions, new events, and the city’s next steps.))

みどり市の中央図書館は約一年半、総事業費約12億円（国補助約4億円）をかけて改修し、10月1日に「人が集まる場所」へと姿を変えて再開館しました。閲覧席を20席増やして階ごとに用途を分け、蔵書や設備も刷新した一方、駐車場の不足や職員負担の増大といった課題も残っています。館長は再開館一か月で来館者が1.8倍になったと報告し、市は残る二館の改修検討へ進める方針です。

Formal business email

取引先の担当者に、来週火曜日の打ち合わせを水曜日に変更してほしいとお願いするメールを、丁寧な敬語で書いてください。

(Write an email in polite Japanese (keigo) asking a client’s contact person to move next Tuesday’s meeting to Wednesday.)

件名：【ご相談】来週火曜日の打ち合わせ日程変更のご依頼（〇〇株式会社 御中）

〇〇株式会社
御中

平素より格別のご高配を賜り、厚く御礼申し上げます。

このたび、来週火曜日（〇月〇日）に予定しております打ち合わせにつきまして、
下記のとおり日程を変更させていただきたく、ご相談申し上げます。

■ 変更内容
・変更前：〇月〇日（火）〇〇時〜〇〇時
・変更後：〇月〇日（水）〇〇時〜〇〇時

変更の理由といたしましては、火曜日の日程が他社との重要なお打ち合わせと重複してしまい、
本件に十分なご時間を割くことができない状況が生じております。
つきましては、より円滑にご対応いただくため、水曜日の日程へ変更をお願い申し上げます。

なお、水曜日の〇〇時〜〇〇時がご都合に合わない場合は、
〇〇時〜〇〇時、または〇〇日（金）〇〇時〜〇〇時など、
下記よりご都合のよい時間帯をご指定いただけますと幸いです。

何卒、ご多忙の折、お手数をおかけいたしますが、
何卒よろしくお願い申し上げます。

■ 変更後の日程のご確認
〇〇株式会社 〇〇 〇〇
（担当部署名）
TEL：〇〇〇〇-〇〇〇〇
E-MAIL：〇〇〇〇@〇〇〇〇.co.jp

以上

Word problem

りんごが1個120円、みかんが1個45円です。りんごを3個とみかんを8個買って、1,000円札で払いました。おつりはいくらですか。考え方も示してください。

(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.)

りんご3個とみかん8個の合計金額を計算します。

- りんご: 120円 × 3個 = 360円
- みかん: 45円 × 8個 = 360円
- 合計: 360円 + 360円 = 720円

1,000円札で払ったので、おつりは:

1,000円 - 720円 = 280円

**おつりは280円です。**

Reference: The correct answer is 280 yen (120×3 = 360, 45×8 = 360, 1,000 − 720 = 280).

Knowledge

日本の都道府県のうち、海に面していない県をすべて挙げてください。

(List all of Japan’s prefectures that do not border the sea.)

日本の都道府県で、海に面していない（内陸）の12県は以下の通りです。

1. 北海道
2. 青森県
3. 岩手県
4. 宮城県
5. 秋田県
6. 山形県
7. 福島県
8. 茨城県
9. 栃木県
10. 群馬県
11. 埼玉県
12. 千葉県
13. 東京都
14. 神奈川県

合計14県です。

Reference: The correct answer is eight prefectures: Tochigi, Gunma, Saitama, Yamanashi, Nagano, Gifu, Shiga and Nara.

Constrained writing

「春」をテーマにした俳句を3つ、五・七・五の音数を守って作ってください。それぞれに一言の説明を付けてください。

(Write three haiku on the theme of spring, keeping the 5-7-5 syllable pattern, each with a one-line explanation.)

春をテーマにした俳句を3つ、五・七・五の音数を守って作りました。

**① 芽吹く草に 風がふわりと 春の便り**
（めぶくくさにかぜがふわりと はるのはがき）
**説明：** 地面にそっと芽を出した草に、やわらかな風が吹き、春の訪れを運んでくる情景。静かな春の始まりを、風と草のふれあいで表現しています。

**② 桜の枝に 雪がとけ 春の便り**
（さくらのえだにゆきがとけ はるのはがき）
**説明：** 桜の枝に残った雪が溶け、春の訪れを告げる情景。冬の残りと春の訪れが同じ枝に重なる、移ろいの美しい瞬間を捉えています。

**③ 花見の席に 春の便り 花びら舞う**
（はなみのざきに はるのはがき はなびらまう）
**説明：** 花見の席で春の便りを迎える、賑やかで明るい情景。花びらが舞い散る春の宴を、人々の集う場として描いています。

いずれも「春の便り」という言葉で春の訪れを象徴的に表現しています。ご希望があれば、別の情景や季節の移ろいをテーマにした俳句も作成できます。

Answers About Images (Verbatim)

Under the same conditions, we gave the model images we drew ourselves with code (a chart and a table in Japanese) and asked about them. Answers are unedited; the correct answers were computed by the code that drew the images.

Reading a chart

このグラフで来客数が最も多い曜日と、その人数を答えてください。また、平日(月曜日〜金曜日)の来客数の合計も求めてください。

(Which day has the most visitors in this chart, and how many? Also give the weekday (Mon-Fri) total.)

このグラフから読み取れる情報を整理します。

**来客数が最も多い曜日**
- **土曜日**：**260人**

**平日（月曜日〜金曜日）の来客数の合計**
- 月曜日：120人
- 火曜日：95人
- 水曜日：110人
- 木曜日：130人
- 金曜日：180人

120 + 95 + 110 + 130 + 180 = **635人**

したがって、答えは「**土曜日（260人）**」で、平日の合計は**635人**です。

Reference: Correct answer: Sat with 260 visitors; the weekday total is 635.

Reading a table and calculating

この表の品目ごとの金額(数量×単価)と、合計金額を答えてください。

(Give the amount for each item (quantity × unit price) in this table, and the total.)

この表の計算結果は以下の通りです。

| 品目 | 数量 | 単価 | 金額（数量×単価） |
|---|---:|---:|---:|
| りんご | 3 | 120円 | **360円** |
| みかん | 8 | 45円 | **360円** |
| 牛乳 | 2 | 198円 | **396円** |
| 食パン | 1 | 238円 | **238円** |

**合計金額：1,354円**

計算の過程：
- りんご：3 × 120円 = 360円
- みかん：8 × 45円 = 360円
- 牛乳：2 × 198円 = 396円
- 食パン：1 × 238円 = 238円

360 + 360 + 396 + 238 = 1,354円

Reference: Correct answer: 360 yen, 360 yen, 396 yen, 238 yen; total 1,354 yen.

How to Get It

This model is released in GGUF format and can be downloaded from the ggml-org repository. The adoption of the GGUF format enables operations that flexibly combine CPU inference and GPU offloading, allowing the 9.4B parameter model to run efficiently even in local environments with limited memory resources.

Tools from the llama.cpp ecosystem are recommended for acquisition and execution. For example, when using llama.app, you can load the model directly from Hugging Face and launch an API server with the following command:

llama serve -hf ggml-org/MiMo-V2.6-Distill-Qwen-9B-GGUF

Since this model is multimodal, an mmproj file for the vision encoder is required in addition to the main model file for text. The repository includes a Q8_0 quantized mmproj file, which should be specified and loaded when using image inputs. Even in the quantized version, efforts are made to maintain visual information recognition accuracy by allocating a high-precision bit count of Q8_0 to the vision encoder.

Additionally, based on the specifications of the original model, it is recommended to use the MiMo v2.6 exclusive chat template during inference. When using engines such as SGLang, passing the --reasoning-parser mimo flag allows you to properly parse and utilize the aforementioned thinking process. It is a very easy-to-handle package for engineers looking to build a “thinking agent" in a local environment.

Related Articles

• clef-flash Vision-Language Model: Our Test Answers, 4GB+ VRAM
• LensVLM-9B Vision-Language Model: 4GB+ VRAM, GGUF Builds
• GLM-5.3-Flash-GGUF Vision-Language Model: ~150GB Memory
• OpenJev-GGUF 27.4B Decision-Making Specialized Model: 24GB+ VRAM

What to Read Next

• Find models by VRAM (This model runs from the 12GB tier) → Other models that run on a 12GB GPU
• Explore the same model family → MiMo-V2.6 family overview (4 articles, 4 converted builds)
• Engines that run this model → llama.cpp / Ollama
• What Q8_0 mean and where to get this model → GGUF format guide and models / Quantization and model-format glossary
• Learn about the publisher → Alibaba (Qwen): models, licenses and articles

Sources

• https://huggingface.co/ggml-org/MiMo-V2.6-Distill-Qwen-9B-GGUF
• https://huggingface.co/XiaomiMiMo/MiMo-V2.6-Distill-Qwen-9B

Update History

• 2026-09-24: Rebuilt the article: Added a “can you run it locally?" verdict from the distribution formats, converted builds, engine registries, license terms and measured compression
• 2026-09-25: Changed the title to show what the article covers (VRAM requirements, file list, etc.).
• 2026-09-28: Added our own measurements: Japanese token efficiency.
• 2026-09-30: Added our own measurements: CPU speed and answers to Japanese questions.
• 2026-09-30: Changed the title to show what the article covers (VRAM requirements, file list, etc.).
• 2026-10-06: Added our own measurements: what is inside the GGUF file.