[whatmodel.sh](../../README.md) / Mistral Small 3.2 24B # Mistral Small 3.2 24B Mistral · released 2025-06 · 24B params · apache-2.0 **score 56.3** ±1.0 · #255 of 732 · from 23 benchmark results across 10 sources Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only. ## What it's good at ● marks Mistral Small 3.2 24B among every model on that task or test, worst to best. 
task                vs the others            rank  counts  tests
agent tasks         ▁▁▃▆●▄▅▅▄▄▂▁     #224 of 313*    6.5%      2
coding                             not ranked yet       —      2
chat & writing      ▁▁▁▃▆▄▆█●▇▇▃     #159 of 402*     14%      2
math                ▁▁▃▅█●▄█▄▂▁▁     #210 of 431*     10%      2
vision & documents  ▁▁▂▂▁▃●▂▆▇█▄      #107 of 153     16%      6
agent tasks · #224 of 313 · 6.5% of its score Beats [o1 (2024-12-17)](o1-2024-12-17.md), [o3-mini](o3-mini.md), [DeepSeek R1 (Jan '25)](deepseek-r1.md) here, though they rank higher overall. Trails [SmolLM3 3B](smollm3-3b.md), [Jamba 1.5 Mini](jamba-1-5-mini.md), [Mellum2 12B-A2.5B Instruct](mellum2-12b-a2-5b-instruct.md) here, though they rank lower overall. 
test                                result  vs the others in its run  difficulty      counts  why
BFCL V4 (December 2025 archive)      37.1%  ▁▁▁▂██●▅▆▂▂▂ beats 41/77  hard · avg 31%    4.6%  at its level
Galileo Agent Leaderboard v2 (act…   26.0%  ▅▃·●▃▅▅█▃█▅▅ beats 3/19   hard · avg 26%    1.9%  at its level
coding · 0.0% of its score 
test                       result  vs the others in its run  counts  why
MBPP+ pass@5 (maker)        78.3%  █··········● beats 1/1         —  not counted yet: no other model in the …
HumanEval+ pass@5 (maker)   92.9%  █··········● beats 1/1         —  not counted yet: no other model in the …
chat & writing · #159 of 402 · 14% of its score Beats [INTELLECT-3](intellect-3.md), [MiniMax-M2](minimax-m2.md), [MiniCPM5 2B](minicpm5-2b.md) here, though they rank higher overall. Trails [LFM2.5 1.2B Thinking](lfm2-5-1-2b-thinking.md), [LFM2.5 8B-A1B](lfm2-5-8b-a1b.md), [LFM2.5 2.6B](lfm2-5-2-6b.md) here, though they rank lower overall. 
test                             result  vs the others in its run    difficulty      counts  why
LMArena (text)                   1356.5  ▁▁▁▂▃▃▃▆●▇█▃ beats 201/385  avg 1260           10%  blunt
Arena-Hard v2 aggregate (maker)   43.1%  ▁▁··▄●▁▃█▁▁▇ beats 5/22     hard · avg 21%    4.0%  sharp · at its level · small run
math · #210 of 431 · 10% of its score Beats [MiniMax-M2.1](minimax-m2-1.md), [DeepSeek V3 (Dec '24)](deepseek-v3.md), [GPT-4o (2024-08-06)](gpt-4o-2024-08-06.md) here, though they rank higher overall. Trails [Granite 8B Code Base (4K)](granite-8b-code-base-4k.md), [Mixtral 8x7B v0.1 (base)](mixtral-8x7b-v0-1.md), [Olmo 3 7B Instruct](olmo-3-7b-instruct.md) here, though they rank lower overall. 
test                               result  vs the others in its run   difficulty            counts  why
OTIS mock AIME 2024–2025            30.3%  ▃▃▄▃▃●▇▇█▃▂▅ beats 65/197  very hard · avg 7.2%    6.1%  sharp · noisy
MATH (subset unspecified) (maker)   69.4%  ▂▄▄▃█●▅▆▄▃▄▁ beats 43/80   medium · avg 60%        4.0%  at its level · small run
vision & documents · #107 of 153 · 16% of its score Beats [DeepSeek V4 Flash Vision (Exp)](deepseek-v4-flash-vision-exp.md), [GPT-4o (2024-08-06)](gpt-4o-2024-08-06.md) here, though they rank higher overall. Trails [Llama 4 Scout (109B-A17B)](llama-4-scout.md) here, though they rank lower overall. 
test                                result  vs the others in its run   difficulty           counts  why
Arena vision (style controlled)     1139.8  ▁▁▂▂▂▂▃●▅▅█▇ beats 42/152  avg 1061               6.6%  blunt
MMMU Pro (maker)                     49.1%  ▂▃▁▃●▇▃▇▇██▅ beats 9/52    hard · avg 41%         3.1%  at its level · small run
MMMU (val) (maker)                   62.5%  ▁····▁▁▃█●▆▅ beats 25/53   medium · avg 55%       2.7%  at its level · small run
VMMU (Vietnamese multimodal under…   31.9%  ▅······●·▅·█ beats 2/7     very hard · avg 36%    2.1%  noisy
DocVQA (maker)                       94.9%  ▁··▃▁▁▁▇●▇▄▅ beats 16/28   very easy · avg 93%    1.0%  blunt · maxed out · small run
ChartQA (maker)                      87.4%  ▁··▁▁▁▁▁▂▅▅● beats 32/44   easy · avg 81%         0.9%  blunt · small run
other tests (12) 
test                                result  vs the others in its run    difficulty            counts  why
GPQA Diamond                         49.1%  ▁▁▁▁▂▄●▄▅▇█▅ beats 67/211   hard · avg 39%           23%  sharp · at its level · several runs
CPTUB average (SpeakLeash, 0-5 ju…   76.5%  ▁▂▂▂▃█▃▆●▄▂▃ beats 74/95    medium · avg 65%        6.2%  at its level
EQ-Bench 3 (emotional intelligenc…  1162.3  ▁·▂▁▄▅●▃█▄▁▁ beats 23/62    avg 1034                6.0%  typical
MMLU (maker)                         80.5%  ▁▁▁▁▁▂▂▅██●▄ beats 169/214  medium · avg 72%        4.5%  small run
MMLU-Pro (maker)                     69.1%  ▁▁·▁▁▁▃▂●▅█▁ beats 83/188   medium · avg 57%        3.6%  at its level · small run
SimpleQA (maker)                     12.1%  ▄▄▄▆█●███·▂█ beats 14/32    very hard · avg 7.4%    3.5%  small run
IFBench (aggregation unspecified)…   34.0%  ▂▆▂●▄▄▂▄▄▆█▇ beats 19/86    hard · avg 29%          2.6%  at its level · small run
EuroEval HellaSwag (English), tes…   90.4%  ▃·▁▁▂▁▂▅█▇●▁ beats 92/100   medium · avg 67%        1.6%  sharp · at its level · noisy
Chess Puzzles (Epoch)                 1.0%  ●▁▁▁▁▂▃▃▃▂▁▁ beats 39/137   very hard · avg 5.8%    1.0%  sharp · too hard for it · noisy
EuroEval Life in the UK (English …   95.1%  ▁▁▁▁▁▁▁▂▄█▄● beats 141/146  easy · avg 82%          0.8%  noisy
GPQA Main (maker)                    44.2%  ●··········█ beats 0/1                                 —  not counted yet: no other model in the …
WildBench v2 (maker)                 65.3%  █··········● beats 1/1                                 —  not counted yet: no other model in the …
## How the score was worked out Each result implies a score on its own, after adjusting for how hard that test is and who ran it. The model's 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 is added up by who ran the tests. [Method](../method.md) 
source                                    results  share of the score
mistralai/Mistral-Small-3.2-24B-Instruc…        9   32% ▇▇▇▇
Epoch AI                                        3   23% ▇▇▇
LMArena leaderboard dataset                     2   16% ▇▇
SpeakLeash CPTUB                                1    6% ▇
EQ-Bench 3 canonical leaderboard results        1    6% ▇
mistralai/Mistral-Small-4-119B-2603 mod…        2    6% ▇
Berkeley Function Calling Leaderboard r…        1    5% ▇
EuroEval leaderboards                           2    2% ▇
VMMU: A Vietnamese Multitask Multimodal…        1    2% ▇
Galileo Agent Leaderboard                       1    2% ▇
prior: pull toward the middle                        0% ▇
weighted average                                   = 56.3
Licences: mistralai/Mistral-Small-3.2-24B-Instruct-2506 model card ([Apache-2.0](../../licenses/mistralai--Mistral-Small-3.2-24B-Instruct-2506.txt)) · Epoch AI ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · SpeakLeash CPTUB ([MIT](../../licenses/hf-leaderboard-spaces-MIT.txt)) · EQ-Bench 3 canonical leaderboard results ([MIT](../../licenses/eqbench3-MIT.txt)) · mistralai/Mistral-Small-4-119B-2603 model card ([Apache-2.0](../../licenses/cards-gap-NOTICE.txt)) · Berkeley Function Calling Leaderboard result archive ([Apache-2.0](../../licenses/bfcl-result-archive-v4-Apache-2.0.txt)) · EuroEval leaderboards ([MIT](https://github.com/EuroEval/leaderboards/blob/1a77f1d2f8ba1ba05ab73287fb090d76bf41b2e0/LICENSE)) · VMMU: A Vietnamese Multitask Multimodal Understanding and Reasoning Benchmark ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)) · Galileo Agent Leaderboard ([Apache-2.0](../../licenses/galileo-agent-leaderboard-Apache-2.0.txt))
Not counted (4) 
source                  test               score  why
mistralai/Mistral-Sma…  GPQA Main          44.2%  too few models have it yet
mistralai/Mistral-Sma…  HumanEval+ pass@5  92.9%  too few models have it yet
mistralai/Mistral-Sma…  MBPP+ pass@5       78.3%  too few models have it yet
mistralai/Mistral-Sma…  WildBench v2       65.3%  too few models have it yet
## How to run it ### Through an API api (DeepInfra) · $0.11/M blended · $0.075 in / $0.2 out per 1M tokens · [deepinfra.com](https://deepinfra.com/mistralai/Mistral-Small-3.2-24B-Instruct-2506) [price source](https://github.com/BerriAI/litellm/blob/c19ce71bcdad234ca87a7af10cee45375ce811a8/model_prices_and_context_window.json): USD per 1M text tokens, standard uncached input and output. Product: Serverless text generation. API model: mistralai/Mistral-Small-3.2-24B-Instruct-2506. Settings: service tier=standard, batch=no, cache=uncached input. Additional/excluded charges: Tools, cache storage, priority service and other modalities excluded.. Currency: USD. · [context source](https://github.com/BerriAI/litellm/blob/c19ce71bcdad234ca87a7af10cee45375ce811a8/model_prices_and_context_window.json): provider maximum input tokens; not an intrinsic model maximum or legacy max\_tokens ### llama.cpp (recommended) 1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md) ```sh # macOS / Linux curl -LsSf https://llama.app/install.sh | sh # Windows irm https://llama.app/install.ps1 | iex ``` 2. Run Mistral Small 3.2 24B (Q4_K_M, needs ~14.3 GB, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/Mistral-Small-3.2-24B-Instruct-2506-GGUF/tree/36798e8b853bcf0ce899aeba9f8a7530ff1991fe)) ```sh llama cli -hf lmstudio-community/Mistral-Small-3.2-24B-Instruct-2506-GGUF:Q4_K_M ``` 3. Or start a local server and open the address it prints (http://localhost:9931) ```sh llama serve -hf lmstudio-community/Mistral-Small-3.2-24B-Instruct-2506-GGUF:Q4_K_M ``` Other versions: Q8_0 (~25.1 GB) `llama cli -hf lmstudio-community/Mistral-Small-3.2-24B-Instruct-2506-GGUF:Q8_0` ### LM Studio 1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download) 2. Run Mistral Small 3.2 24B (Q4_K_M, needs ~14.3 GB, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/Mistral-Small-3.2-24B-Instruct-2506-GGUF/tree/36798e8b853bcf0ce899aeba9f8a7530ff1991fe)) ```sh lms get https://huggingface.co/lmstudio-community/Mistral-Small-3.2-24B-Instruct-2506-GGUF@Q4_K_M -y ``` 3. Then run lms ls to see its name, and lms chat followed by that name Other versions: Q8_0 (~25.1 GB) `lms get https://huggingface.co/lmstudio-community/Mistral-Small-3.2-24B-Instruct-2506-GGUF@Q8_0 -y` · 4bit (~13.5 GB, Mac only) `lms get https://huggingface.co/lmstudio-community/Mistral-Small-3.2-24B-Instruct-2506-MLX-4bit -y` · 8bit (~25.9 GB, Mac only) `lms get https://huggingface.co/lmstudio-community/Mistral-Small-3.2-24B-Instruct-2506-MLX-8bit -y` ### Ollama 1. Install Ollama. [Download it (macOS / Windows)](https://ollama.com/download) ```sh # Linux curl -fsSL https://ollama.com/install.sh | sh ``` 2. Run Mistral Small 3.2 24B (Q4_K_M, needs ~14.9 GB, from [ollama.com](https://ollama.com/library/mistral-small3.2/tags)) ```sh ollama run mistral-small3.2:24b ``` Other versions: Q8_0 (~29.3 GB) `ollama run mistral-small3.2:24b-instruct-2506-q8_0` ### MLX (Mac) 1. Install MLX (Mac). Apple silicon Macs only, text only. ```sh # with uv uv tool install mlx-lm # or pip install mlx-lm ``` 2. Run Mistral Small 3.2 24B (4bit, needs ~13.5 GB, Mac only, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/Mistral-Small-3.2-24B-Instruct-2506-MLX-4bit/tree/61d29b9c1c12db05b4d156ea4bd743e9f00be8e6)) ```sh mlx_lm.chat --model lmstudio-community/Mistral-Small-3.2-24B-Instruct-2506-MLX-4bit ``` 3. Or start an OpenAI-compatible server at http://localhost:8080/v1 ```sh mlx_lm.server --model lmstudio-community/Mistral-Small-3.2-24B-Instruct-2506-MLX-4bit ``` Other versions: 8bit (~25.9 GB, Mac only) `mlx_lm.chat --model lmstudio-community/Mistral-Small-3.2-24B-Instruct-2506-MLX-8bit` Model repository and documentation: [huggingface.co/mistralai](https://huggingface.co/mistralai/Mistral-Small-3.2-24B-Instruct-2506) ## Quantization evidence How much of the full-precision result a quantized version kept, measured within one study. Rows from the same study share a table and aren't independent. They don't describe every download of that precision. 
precision                       test                                full → quantized    kept
NVFP4 (FP4 weights and activa…  IFEval 0-shot (OpenLLM v2 config)      0.841 → 0.844  100.4%
NVFP4 (FP4 weights and activa…  OpenLLM v1 average of 6 tasks (AR…     0.782 → 0.768   98.1%
NVFP4 (FP4 weights and activa…  MMLU (OpenLLM v1 config, lm-evalu…     0.811 → 0.794   98.0%
NVFP4 (FP4 weights and activa…  OpenLLM v2 average of 6 tasks (IF…     0.525 → 0.512   97.7%
NVFP4 (FP4 weights and activa…  GSM8K (OpenLLM v1 config, lm-eval…     0.896 → 0.871   97.2%
NVFP4 (FP4 weights and activa…  MMLU-Pro 5-shot (OpenLLM v2 confi…     0.508 → 0.484   95.3%
Each kept value links to its source. [interactive version](https://whatmodel.sh/model/mistral-small-3-2) · [JSON](https://whatmodel.sh/api?format=json&all=1)