MiniMax M1 40k — benchmark results
MiniMax M1 open hybrid-attention reasoning model at the 40K thinking budget. Provider: MiniMax. Released 2025-06-01. Access: Open.
Unified ELO 1571 ± 48, rank #562 of 1776 rated models, from 26 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| LLM Stats (OpenAI-MRCR: 2 needle 1M) | 58.6 | Score (%) | 100 |
| AA MATH-500 | 97.2 | Accuracy (%) | 87.6 |
| LLM Stats (OpenAI-MRCR: 2 needle 128k) | 76.1 | Score (%) | 87.5 |
| SpeechMap Compliance | 87.2 | % Requests Completed | 84.7 |
| AA LiveCodeBench | 65.71 | Pass@1 (%) | 73.8 |
| AA MMLU-Pro | 80.83 | Accuracy (%) | 71.5 |
| LLM Stats (LongBench v2) | 61 | Score (%) | 66.7 |
| AA-LCR | 51.7 | Score (self-reported) | 66.1 |
| AA SciCode | 37.85 | Accuracy (%) | 65 |
| LLM Stats (AIME 2024) | 83.3 | Score (%) | 60.4 |
| AA Long Context Reasoning | 51.67 | Accuracy (%) | 60.1 |
| AA Humanity's Last Exam | 7.51 | Accuracy (%) | 53.4 |
Interactive version: theaggregate.ai/model?slug=minimax-m1-40k · How the rankings work · Data refreshed daily, snapshot 2026-07-22.