whatmodel.sh / Mistral Small Instruct 22B (2409)

Mistral Small Instruct 22B (2409)

+ contribute
Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Mistral Small Instruct 22B (2409) is an open-weights language model from Mistral AI, released September 2024. Its published size is 22.25B parameters and a 32k-token context window. Licence: mistral-research.

Catalogue identity notes

Maker repo mistralai/Mistral-Small-Instruct-2409 @4600506: licence tag other (mrl). 22,247,282,688 parameters from the Hugging Face safetensors metadata. Dense, so active = total. Released 2024-09: month the maker's Hugging Face repo was created (createdAt 2024-09-17). Ollama tags mistral-small:22b-instruct-2409-q4_K_M, mistral-small:22b-instruct-2409-q8_0 are the library's instruct/chat builds of this checkpoint (tag names on the pinned tags page); download sizes are estimated from the parameter count.

49.4±1.9#354 of 719 LLMsfrom 9 benchmark results across 6 sources

$ how it compares

The LLMs ranked either side of Mistral Small Instruct 22B (2409), out of 828 with a score, then more from Mistral AI.

#modelscorewhereRAMreleased
423Nova Pro Amazon50±1api $1.40/M2024-12
424WizardCoder 33B V1.1 WizardLM50±1local~22GB2024-01
425Mistral Small Instruct 22B (2409) Mistral AI49±2local~15GB2024-09
426Qwen2-Math 72B Instruct Alibaba49±1local~32GB2024-08
427OLMo 2 32B Instruct (0325) Ai249±1local~21GB2025-03
more from Mistral AI
419Codestral 22B v0.1 Mistral AI50±1local~15GB2024-05
587Mathstral 7B v0.1 Mistral AI42±1local~5.8GB2024-07
612Mixtral 8x7B v0.1 (base) Mistral AI41±1local~29GB2023-12

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #255 of 300 * · 6.1% of its score · 1 test

Beats DeepSeek R1 0528 Qwen3 8B, Magistral Small 1.0 (2506), Llama 4 Scout (109B-A17B) here, though they rank higher overall. Trails Granite 4.0 H 1B, Granite 4.0 1B, Qwen2.5 3B Instruct here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
Galileo Agent Leaderboard v1 (tool selection quality)75.0% ✎▃▃▃▃▆▅▆▅█▂▅▂ beats 5/31easy · avg 79%▇▇ 6.1%blunt
coding▁▁▁▃▆██▇▆▄▃▁ #239 of 458 · 26% of its score · 4 tests

Beats Qwen3-VL 32B Instruct, Solar Open 100B (102B-A12B), GLM-4.7-Flash here, though they rank higher overall. Trails QwQ-32B-Preview, InternLM2.5 20B Chat here, though they rank lower overall.

testresultvs the others in its rundifficultycountswhy
CodeReviewQA: change localisation (hard), invariant accuracy31.6% ✎█▂▁▁▂▂▂▃▅▅▅▇ beats 27/47medium · avg 36%▇▇▇▇ 10%sharp · at its level · noisy
CodeReviewQA: solution identification (hard), invariant accuracy60.1% ✎▃··▁▂▂▃▂█▃▄▂ beats 34/47medium · avg 54%▇▇▇▇ 8.8%sharp · at its level · noisy
CodeReviewQA: automated code refinement, exact match rate45.6% ✎▁▁▁▁▁▁▂▂▃█▅▄ beats 43/47hard · avg 33%▇▇▇ 6.7%at its level
CodeReviewQA: change type recognition, invariant accuracy76.7% ✎▁▁▁······▁▃█ beats 36/47easy · avg 67%▇ 0.1%blunt · at its level · noisy
other tests4 not in a task · 67% of its score
testresultvs the others in its rundifficultycountswhy
EvalPerf correctness (118 tasks, 20240328)69.5% ✎▂·▂▂▂▆▅▆▆▆▃█ beats 6/29easy · avg 76%▇▇▇▇▇▇▇▇ 20%typical
Polish EQ-Bench v2 (SpeakLeash)72.9% ✎▁··▁▁▂▂▆▅▆█▆ beats 62/70medium · avg 64%▇▇▇▇▇▇▇ 17%at its level
BRIDGE clinical text, zero-shot average (87 tasks)35.2% ✎▁▁▁▁▁▂▂▃▄▆█▃ beats 46/97hard · avg 32%▇▇▇▇▇▇ 15%blunt · at its level
Open Persian LLM Leaderboard v2 average (21 tasks)44.0% ✎▂▄▂▃█▄▃▆▃▅▂▂ beats 17/81medium · avg 58%▇▇▇▇▇ 13%blunt · at its level

$ where to run it

llama.cpp · recommended

version

  1. Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
    macOS / Linux
    curl -LsSf https://llama.app/install.sh | sh
    Windows
    irm https://llama.app/install.ps1 | iex

    other ways to install ↗

  2. Run Mistral Small Instruct 22B (2409) · needs ~13.3 GB · from huggingface.co/lmstudio-community
    chat in the terminal
    llama cli -hf lmstudio-community/Mistral-Small-Instruct-2409-GGUF:Q4_K_M
  3. Or start a local server and open the address it prints (http://localhost:9931)
    llama serve -hf lmstudio-community/Mistral-Small-Instruct-2409-GGUF:Q4_K_M

LM Studio

version

  1. Install LM Studio · Open the app once before using the lms command.

    download it (macOS / Windows / Linux) ↗

  2. Run Mistral Small Instruct 22B (2409) · needs ~13.3 GB · from huggingface.co/lmstudio-community

    open in LM Studio ↗

  3. Or download it in the terminal
    lms get https://huggingface.co/lmstudio-community/Mistral-Small-Instruct-2409-GGUF@Q4_K_M -y
  4. Then run lms ls to see its name, and lms chat followed by that name

Ollama

version

  1. Install Ollama

    download it (macOS / Windows) ↗

    Linux
    curl -fsSL https://ollama.com/install.sh | sh
  2. Run Mistral Small Instruct 22B (2409) · needs ~13.85 GB · from ollama.com
    chat in the terminal
    ollama run mistral-small:22b-instruct-2409-q4_K_M

MLX (Mac)

version

  1. Install MLX (Mac) · Apple silicon Macs only, text only.
    with uv
    uv tool install mlx-lm
    or
    pip install mlx-lm
  2. Run Mistral Small Instruct 22B (2409) · needs ~12.5 GB, Mac only · from huggingface.co/mlx-community
    chat in the terminal
    mlx_lm.chat --model mlx-community/Mistral-Small-Instruct-2409-4bit
  3. Or start an OpenAI-compatible server at http://localhost:8080/v1
    mlx_lm.server --model mlx-community/Mistral-Small-Instruct-2409-4bit

Official weights and documentation: huggingface.co/mistralai ↗

$ how the score was worked out · CodeReviewQA: The Code Review… 26%, EvalPlus / EvalPerf published… 20%, SpeakLeash Polish EQ-Bench le… 17% and 3 more

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its counts column above; here it's added up by who ran the tests. How it works

sourceresultsshare of the scorelicence
CodeReviewQA: The Code Review Comprehension Assessment for Large Language Models (arXiv:2503.16167v2)4▇▇▇▇▇▇▇▇ 26%CC-BY-4.0
EvalPlus / EvalPerf published evaluation runs1▇▇▇▇▇▇ 20%Apache-2.0
SpeakLeash Polish EQ-Bench leaderboard1▇▇▇▇▇ 17%Apache-2.0
BRIDGE Medical Leaderboard (Harvard Medical School / Brigham and Women's YLab)1▇▇▇▇▇ 15%MIT
Open Persian LLM Leaderboard v2 (Part DP AI and AUT NLP Lab)1▇▇▇▇ 13%Apache-2.0
Galileo Agent Leaderboard (v1 and v2 results)1▇▇ 6%Apache-2.0
prior · pull toward the middle▇ 2%
weighted average= 49.4

$ improve this page

Update model informationAdd a place to run itAdd a test resultPropose a benchmarkCorrect a benchmark scoreEdit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated · sources and licences · json