[whatmodel.sh](../../README.md) / MiniMax-M3
# MiniMax-M3
MiniMax · released 2026-06 · 428B params (23B active) · minimax-community
**score 68.9** ±0.9 · #88 of 732 · from 16 benchmark results across 8 sources
Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only.
## What it's good at
● marks MiniMax-M3 among every model on that task or test, worst to best.
task vs the others rank counts tests
agent tasks ▁▁▃▆█▄▅▅●▄▂▁ #73 of 313 37% 5
coding ▁▁▁▃▅▇█▇▅●▃▁ #60 of 469* 20% 2
chat & writing ▁▁▁▃▆▄▆█▆▇●▃ #68 of 402* 8.6% 1
math ▁▁▃▅█▆▄●▄▂▁▁ #143 of 431* 3.3% 1
vision & documents ▁▁▂▂▁▃▄▂▆●█▄ #65 of 153 8.8% 3
web search ▁▁▁▅●█▃▄·▁·▁ #47 of 84* 2.4% 1
agent tasks · #73 of 313 · 37% of its score
Beats [GPT-5 Pro](gpt-5-pro.md), [GPT-5](gpt-5.md), [Qwen3.6 Max Preview](qwen3-6-max-preview.md) here, though they rank higher overall. Trails [MiMo-V2.5 (310B-A15B)](mimo-v2-5.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
Arena Agent (October 2026 aggrega… -0.1016 IPS ▁▁▁●▃▂█▁▂▂▁▁ beats 5/59 avg -0.379 IPS 22% sharp
MCP-Atlas (public set) 74.2% █··█·█●·████ beats 3/7 hard · avg 30% 5.2% at its level
GameLogicBench solve rate (72 God… 19.4% █●·█·▄·██▄·▄ beats 2/11 very hard · avg 5.1% 4.9% typical
EEGAgentBench overall (tool-based… 53.0% ▂·▁▁··▇▇█●·▁ beats 24/25 hard · avg 28% 2.6% blunt · at its level
Legal Research Bench (agentic, al… 29.8% ▄█··▄▄▄●█·▄█ beats 7/12 very hard · avg 6.3% 2.4% typical
coding · #60 of 469 · 20% of its score
Beats [GPT-5.4](gpt-5-4.md), [Gemini 3.1 Pro (Preview)](gemini-3-1-pro.md), [DeepSeek V4 Flash Vision (Exp)](deepseek-v4-flash-vision-exp.md) here, though they rank higher overall.
test result vs the others in its run difficulty counts why
Arena WebDev (Code Arena) 1483.3 ▁▁▃▂▅█●▆▅▅▁▂ beats 70/129 avg 981 15% sharp
GameLogicBench solve rate (72 God… 19.4% █●·█·▄·██▄·▄ beats 2/11 very hard · avg 5.1% 4.9% typical
chat & writing · #68 of 402 · 8.6% of its score
Beats [Claude Haiku 5.5](claude-haiku-5-5.md), [DeepSeek V4 Flash (0731)](deepseek-v4-flash.md), [GPT-5.2](gpt-5-2.md) here, though they rank higher overall. Trails [Nemotron Cascade 2 30B-A3B](nemotron-cascade-2-30b-a3b.md), [Gemma 4 31B](gemma-4-31b.md), [Gemini 2.5 Pro](gemini-2-5-pro.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
LMArena (text) 1439.2 ▁▁▁▂▃▃▃▆▆▇●▃ beats 302/385 avg 1260 8.6% blunt
math · #143 of 431 · 3.3% of its score
Beats [Claude Sonnet 4.6](claude-sonnet-4-6.md) here, though they rank higher overall. Trails [Olmo 3 32B Think](olmo-3-32b-think.md), [Nemotron-Cascade-8B](nemotron-cascade-8b.md), [Olmo 3.1 32B Think](olmo-3-1-32b-think.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
OTIS mock AIME 2024–2025 71.1% ▃▃▄▃▃▃▇●█▃▂▅ beats 105/197 very hard · avg 7.2% 3.3% sharp · noisy
vision & documents · #65 of 153 · 8.8% of its score
Beats [DeepSeek V4 Flash Vision (Exp)](deepseek-v4-flash-vision-exp.md), [GPT-5](gpt-5.md) here, though they rank higher overall. Trails [Gemma 4 26B-A4B](gemma-4-26b-a4b.md), [Qwen3.6 Flash](qwen3-6-flash.md), [Qwen3.5 35B-A3B](qwen3-5-35b-a3b.md) here, though they rank lower overall.
test result vs the others in its run difficulty counts why
Arena vision (style controlled) 1233.3 ▁▁▂▂▂▂▃▄▅●█▇ beats 91/152 avg 1061 5.7% blunt
Arena document (style controlled) 1433.0 ▁▁▃●▄▃▇█▅▇▄▅ beats 5/37 avg 1331 3.0% blunt
PureDocBench document parsing (Av… 50.7% ●·▁·▂▂▅█·▁▂▂ beats 1/22 medium · avg 62% 0.2% blunt · at its level
web search · #47 of 84 · 2.4% of its score
Beats [Gemini 3.1 Pro (Preview)](gemini-3-1-pro.md), [Qwen3.7 Max](qwen3-7-max.md), [GPT-5.2](gpt-5-2.md) here, though they rank higher overall. Trails [GPT-4o mini (2024-07-18)](gpt-4o-mini-2024-07-18.md), [Mistral Small 3.1 24B (2503)](mistral-small-3-1.md), [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
Legal Research Bench (agentic, al… 29.8% ▄█··▄▄▄●█·▄█ beats 7/12 very hard · avg 6.3% 2.4% typical
other tests (5)
test result vs the others in its run difficulty counts why
GPQA Diamond 90.9% ▁▁▁▁▂▄▆▄▅▇●▅ beats 181/211 hard · avg 39% 10% sharp
Mystery Game Puzzles (Epoch) 7.0% ●··▂▃▇▄█▂▄▂▁ beats 13/77 very hard · avg 9.4% 6.1% sharp · noisy
Chess Puzzles (Epoch) 14.0% █▁▁▁▁●▃▃▃▂▁▁ beats 67/137 very hard · avg 5.8% 4.9% sharp · noisy
Icelandic LLM Leaderboard average… 31.3% ▁▁▃▂▂▄●▃▄▄▄█ beats 13/38 very hard · avg 14% 3.1% at its level
Legal Research Bench (single-shot… 2.9% ▃▃●▃·▅···█▃▃ beats 3/12 very hard · avg 0.8% 2.6% too hard for it
source results share of the score
LMArena leaderboard dataset 5 54% ▇▇▇▇▇▇▇
Epoch AI 4 24% ▇▇▇
zai-org/GLM-5.2 model card 1 5% ▇
Legal Research Bench 2 5% ▇
GameLogicBench: Evaluating Coding Agent… 1 5% ▇
Miðeind Icelandic LLM Leaderboard 1 3% ▇
EEGAgentBench: Benchmarking LLM Agents … 1 3% ▇
How Far Is Document Parsing from Solved… 1 0% ▇
prior: pull toward the middle 0% ▇
weighted average = 68.9
Licences: LMArena leaderboard dataset ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Epoch AI ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · zai-org/GLM-5.2 model card ([mit](../../licenses/zai-org--GLM-5.2.txt)) · Legal Research Bench ([CC-BY-4.0](https://creativecommons.org/licenses/by/4.0/)) · GameLogicBench: Evaluating Coding Agents on Runtime Game Logic with Tick-Level State Assertions ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · Miðeind Icelandic LLM Leaderboard ([CC BY 4.0](../../licenses/mideind-icelandic-llm-leaderboard-CC-BY-4.0.txt)) · EEGAgentBench: Benchmarking LLM Agents on Short- and Long-Horizon EEG Analysis ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)) · How Far Is Document Parsing from Solved? PureDocBench: A Source-Traceable Benchmark across Clean ([CC BY 4.0](https://creativecommons.org/licenses/by/4.0/))
## How to run it
### Through an API
api (Deep Infra) · $0.49/M blended · $0.28 in / $1.1 out per 1M tokens · [deepinfra.com](https://deepinfra.com/models)
[price source](https://github.com/anomalyco/models.dev/blob/a668908cfe7073faa0363699486fc41a3ff52412/providers/deepinfra/models/MiniMaxAI/MiniMax-M3.toml): Standard base token rate from models.dev. Provider-specific tiers, long-context surcharges, caching and promotions may differ. · [context source](https://github.com/anomalyco/models.dev/blob/a668908cfe7073faa0363699486fc41a3ff52412/models/minimax/MiniMax-M3.toml)
### 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 MiniMax-M3 (ud-q4_k_m, needs ~257.3 GB, from [huggingface.co/unsloth](https://huggingface.co/unsloth/MiniMax-M3-GGUF/tree/main))
```sh
llama cli -hf unsloth/MiniMax-M3-GGUF:UD-Q4_K_M
```
3. Or start a local server and open the address it prints (http://localhost:9931)
```sh
llama serve -hf unsloth/MiniMax-M3-GGUF:UD-Q4_K_M
```
Other versions: ud-q2_k_xl (~128.9 GB) `llama cli -hf unsloth/MiniMax-M3-GGUF:UD-Q2_K_XL`
### 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 MiniMax-M3 (ud-q4_k_m, needs ~257.3 GB, from [huggingface.co/unsloth](https://huggingface.co/unsloth/MiniMax-M3-GGUF/tree/main))
```sh
lms get https://huggingface.co/unsloth/MiniMax-M3-GGUF@UD-Q4_K_M -y
```
3. Then run lms ls to see its name, and lms chat followed by that name
Other versions: ud-q2_k_xl (~128.9 GB) `lms get https://huggingface.co/unsloth/MiniMax-M3-GGUF@UD-Q2_K_XL -y` · 4bit (~241.5 GB, Mac only) `lms get https://huggingface.co/mlx-community/MiniMax-M3-4bit -y` · 8bit (~454.4 GB, Mac only) `lms get https://huggingface.co/inferencerlabs/MiniMax-M3-MLX-Q8 -y`
### 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 MiniMax-M3 (4bit, needs ~241.5 GB, Mac only, from [huggingface.co/mlx-community](https://huggingface.co/mlx-community/MiniMax-M3-4bit/tree/29909f4575501289881466f3b3bd910fdc95bc7c))
```sh
mlx_lm.chat --model mlx-community/MiniMax-M3-4bit
```
3. Or start an OpenAI-compatible server at http://localhost:8080/v1
```sh
mlx_lm.server --model mlx-community/MiniMax-M3-4bit
```
Other versions: 8bit (~454.4 GB, Mac only) `mlx_lm.chat --model inferencerlabs/MiniMax-M3-MLX-Q8`
Model repository and documentation: [huggingface.co/MiniMaxAI](https://huggingface.co/MiniMaxAI/MiniMax-M3)
[interactive version](https://whatmodel.sh/model/minimax-m3) · [JSON](https://whatmodel.sh/api?format=json&all=1)