Codesota · Models · MiniMax M2.5MiniMax5 results · 3 benchmarks
Model card

MiniMax M2.5.

MiniMaxopen-source229B params
§ 01 · Benchmarks

Every benchmark MiniMax M2.5 has a recorded score for.

#BenchmarkArea · TaskMetricValueRankDateSource
01SWE-BenchComputer Code · Code Generationresolve-rate-agentic80.2%#3/252026-01-01source ↗
02SWE-BenchComputer Code · Code Generationresolve-rate80.2%#4/322026-01-01source ↗
03SWE-BenchComputer Code · Code Generationresolve-rate80.2%#4/322026-01-01source ↗
04SWE-Bench VerifiedComputer Code · Code Generationresolve-rate80.2%#5/392026-02-12source ↗
05SWE-bench VerifiedAgentic AI · SWE-benchresolve-rate80.2%#5/81source ↗
Rank column shows this model’s position vs all other models scored on the same benchmark + metric (competitors after the slash). #1 in red means current SOTA. Sorted by rank, then newest result.
§ 02 · Strengths by area

Where MiniMax M2.5 actually performs.

Computer Code
2
benchmarks
avg rank #4.0
Agentic AI
1
benchmark
avg rank #5.0
§ 03 · Papers

1 paper with results for MiniMax M2.5.

  1. 2023-10-10· Computer Code· 2 results

    SWE-bench: Can Language Models Resolve Real-World GitHub Issues?

    Carlos E. Jimenez, John Yang, Alexander Wettig, Shunyu Yao et al.
§ 04 · Related models

Other MiniMax models scored on Codesota.

MiniMax M2.1
3 results
MiniMax M2
2 results
MiniMax M1 40K
1 result
MiniMax M1 80K
1 result
MiniMax M2.5
Unknown params · 1 result
§ 05 · Sources & freshness

Where these numbers come from.

minimax-agent
1
result
sota-timeline
1
result
swebench-leaderboard
1
result
minimax-blog
1
result
editorial
1
result
5 of 5 rows marked verified. · first result 2026-01-01, latest 2026-02-12.