Codesota · Models · Kimi K2.5Moonshot AI5 results · 4 benchmarks
Model card

Kimi K2.5.

Moonshot AIapiUndisclosed params
§ 01 · Benchmarks

Every benchmark Kimi K2.5 has a recorded score for.

#BenchmarkArea · TaskMetricValueRankDateSource
01SWE-BenchComputer Code · Code Generationresolve-rate-agentic76.8%#10/252026-01-01unverified
02SWE-Bench VerifiedComputer Code · Code Generationresolve-rate76.8%#10/39source ↗
03MMLU-ProReasoning · Commonsense Reasoningaccuracy87.1%#11/202026-04-20source ↗
04SWE-BenchComputer Code · Code Generationresolve-rate76.8%#12/322026-01-01source ↗
05SWE-bench VerifiedAgentic AI · SWE-benchresolve-rate76.8%#13/81source ↗
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 Kimi K2.5 actually performs.

Computer Code
2
benchmarks
avg rank #10.7
Reasoning
1
benchmark
avg rank #11.0
Agentic AI
1
benchmark
avg rank #13.0
§ 03 · Papers

1 paper with results for Kimi K2.5.

  1. 2023-10-10· Computer Code· 1 result

    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 Moonshot AI models scored on Codesota.

Kimi K2-Thinking-0905
2 results
Kimi K2-Instruct-0905
1 result
Kimi K2.5
1 result
Kimi k1.5
1 result
Kimi-Audio
0 results
§ 05 · Sources & freshness

Where these numbers come from.

moonshot-agent
1
result
moonshot-blog
1
result
llm-stats
1
result
swebench-leaderboard
1
result
editorial
1
result
2 of 5 rows marked verified. · first result 2026-01-01, latest 2026-04-20.