whatmodel.sh / MiniMax-M1 (40k)
MiniMax-M1 (40k)
↗ share
+ contribute
MiniMax-M1 (40k) is an open-weights language model from MiniMax, released June 2025. Its published size is 456B parameters (45.9B active per token). Licence: apache-2.0.
Catalogue identity notes
Comparison-only, proposed so the MiniMax-M1-80k card runs count. The pinned MiniMax-M1-80k card (@8d1494b) evaluates two own checkpoints side by side, 'MiniMax-M1-80K' (catalogue minimax-m1) and 'MiniMax-M1-40K' (40K thinking budget). Weights: https://huggingface.co/MiniMaxAI/MiniMax-M1-40k (sha 2d1d1c2f00c97fc1245bfce7648649b76e0a8e6e, created 2025-06-05, apache-2.0, 456.1B params in safetensors metadata). Same architecture as M1-80k (456B total, 45.9B active per the card); differs only in RL thinking budget.
$ how it compares
The LLMs ranked either side of MiniMax-M1 (40k), out of 828 with a score, then more from MiniMax.
| # | model | score | where | ||
|---|---|---|---|---|---|
| 215 |  Gemini 2.5 Flash-Lite Preview (06-17) | 62±2 | comparison only | ||
| 216 | Gemini 2.5 Flash | 62±1 | api $0.85/M | ||
| 217 | MiniMax-M1 (40k) | 61±1 | comparison only | ||
| 218 |  Grok 4 Fast | 61±2 | comparison only | ||
| 219 |  GLM-4.5V | 61±2 | api $0.90/M +local | ||
| more from MiniMax | |||||
| 84 | MiniMax-M3 | 70±1 | api $0.48/M +local | ||
| 120 | MiniMax-M2.7 | 68±1 | api $0.53/M +local | ||
| 150 | MiniMax-M2.5 | 66±1 | api $0.53/M +local | ||
$ what it's good at
agent tasks█··········█ #2 of 2 * · 10% of its score · 3 tests
| test | result | vs the others in its run | counts |
|---|---|---|---|
| SWE-bench Verified lab | 55.6% ✎ | ▁▁▁▁▁▁▂▄█▆▁▁ beats 43/131 | 10% |
| τ-bench Airline (GPT-4.1 user) lab | 60.0% ✎ | █··········█ beats 0/1 | — |
| τ-bench Retail (GPT-4.1 user) lab | 67.8% ✎ | █··········█ beats 1/1 | — |
coding█··········█ #2 of 2 * · 15% of its score · 2 tests
| test | result | vs the others in its run | counts |
|---|---|---|---|
| SWE-bench Verified lab | 55.6% ✎ | ▁▁▁▁▁▁▂▄█▆▁▁ beats 43/131 | 10% |
| LiveCodeBench 2024-08–2025-05 (version unstated) lab | 62.3% ✎ | ▃···█··▅··▃▃ beats 1/7 | 5.1% |
math█··········█ #2 of 2 · 31% of its score · 3 tests
other tests7 not in a task · 53% of its score
| test | result | vs the others in its run | counts |
|---|---|---|---|
| GPQA Diamond lab | 69.2% ✎ | ▁▁▁▁▂▄▅▅███▄ beats 149/310 | 14% |
| LongBench v2 lab | 61.0% ✎ | ▂·▁·▂▂▁▃▄▁█▄ beats 23/28 | 7.6% |
| Humanity's Last Exam lab | 7.2% ✎ | ▁▁▁▃▆▆▇██▆▁▃ beats 8/104 | 7.2% |
| MMLU-Pro lab | 80.6% ✎ | ▁▁·▁▁▁▃▂▄▅█▁ beats 128/188 | 6.8% |
| ZebraLogic lab | 80.1% ✎ | █·▁▁▁▇·▄▄▁▄▁ beats 11/24 | 6.6% |
| SimpleQA lab | 17.9% ✎ | ▄▄▄▆█▆███·▂█ beats 18/32 | 6.5% |
| MultiChallenge lab | 44.7% ✎ | ▁·▁·▁▃▆▅▃▁█▅ beats 8/29 | 4.4% |
$ where to run it
No verified API or download recorded yet · add one
$ how the score was worked out · MiniMaxAI/MiniMax-M1-80k mode… 99%
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
| source | results | share of the score |
|---|---|---|
 MiniMaxAI/MiniMax-M1-80k model card · lab-reported | 12 | 99% |
| prior · pull toward the middle | 1% | |
| weighted average | = 61.4 |
not counted (2)
| MiniMaxAI/MiniMax-M1-80k model card | τ-bench Airline (GPT-4.1 user) [correct] | 60.0% | fewer than three models have this test yet, or this source ran it for this model alone |
| MiniMaxAI/MiniMax-M1-80k model card | τ-bench Retail (GPT-4.1 user) [correct] | 67.8% | fewer than three models have this test yet, or this source ran it for this model alone |
$ improve this page
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