Frontier Log

MiniMaxModel

MiniMax M2.5

Made by
MiniMax
Released
Announced in
MiniMax M2.5: Built for Real-World Productivity.minimax.io
Event
MiniMax M2.5
API record
/api/v1/models/minimax-m2-5

Benchmark results

BenchmarkScoreSource check
Humanity's Last ExamNo tools · in-house run with Artificial Analysis Intelligence Index methods; full or text-only set not stated19.4%
CheckedA person checked this number against its source on 20 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
SWE-bench Verified80.2%
CheckedA person checked this number against its source on 17 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
SWE-bench ProClaude Code harness; average of 4 runs · The source doesn't say whether this is the public or the commercial set55.4%
CheckedA person checked this number against its source on 17 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
CorroboratedThe same number is printed in 1 other lab document: 1
GPQA Diamondin-house run with Artificial Analysis Intelligence Index methods85.2%
CheckedA person checked this number against its source on 20 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
BrowseCompWebExplorer agent; with context management: all history discarded past 30% of the context76.3%
CheckedA person checked this number against its source on 17 Oct 2026. This confirms the number matches what the lab published, not that the score was reproduced.
Sources

Reported by MiniMax

Events

Stories

All models