MiniMax M1 80k: benchmark results

MiniMax M1 open hybrid-attention reasoning model at the 80K thinking budget. Provider: MiniMax. Released 2025-06-01. Access: Open.

Unified ELO 1557 ± 1, rank #414 of 1392 rated models, from 30 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
BenCzechMark71.78Average Score (%)86.3
LLM Stats (LongBench v2)61.5Score (%)75
LLM Stats (OpenAI-MRCR: 2 needle 128k)73.4Score (%)75
LLM Stats (OpenAI-MRCR: 2 needle 1M)56.2Score (%)75
ZeroEval MATH-50096.8MATH-500 Score69.4
LLM Stats (AIME 2024)86Score (%)69.2
AA Omniscience - Science, Engineering & Mathematics31.5Accuracy (%)62.3
AA Omniscience - Health23.12Accuracy (%)60.4
BenchmarkList ECI118.86Capability Index (ECI)60
AA Omniscience - Business18.72Accuracy (%)57.9
AA Omniscience - Software Engineering (SWE)28.8Accuracy (%)56.2
AA-Omniscience Accuracy22.47Accuracy (%)55.2

Interactive version: theaggregate.ai/model?slug=minimax-m1-80k · How It Works · Data refreshed daily, snapshot 2026-09-05.