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

BenchmarkScoreMetricPercentile
LLM Stats (OpenAI-MRCR: 2 needle 1M)58.6Score (%)100
AA MATH-50097.2Accuracy (%)87.6
LLM Stats (OpenAI-MRCR: 2 needle 128k)76.1Score (%)87.5
SpeechMap Compliance87.2% Requests Completed84.7
AA LiveCodeBench65.71Pass@1 (%)73.8
AA MMLU-Pro80.83Accuracy (%)71.5
LLM Stats (LongBench v2)61Score (%)66.7
AA-LCR51.7Score (self-reported)66.1
AA SciCode37.85Accuracy (%)65
LLM Stats (AIME 2024)83.3Score (%)60.4
AA Long Context Reasoning51.67Accuracy (%)60.1
AA Humanity's Last Exam7.51Accuracy (%)53.4

Interactive version: theaggregate.ai/model?slug=minimax-m1-40k · How the rankings work · Data refreshed daily, snapshot 2026-07-22.