Model card
Kimi-K2.5.
Moonshot.AIopen-source
§ 02 · Benchmarks
Every benchmark Kimi-K2.5 has a recorded score for.
| # | Benchmark | Area · Task | Metric | Value | Rank | Date | Source |
|---|---|---|---|---|---|---|---|
| 01 | AIME 2025 | Reasoning · Mathematical Reasoning | accuracy | 96.1% | #3 | — | source ↗ |
| 02 | Video-MME | Multimodal · Video Understanding | accuracy | 87.4% | #3 | — | source ↗ |
| 03 | LiveCodeBench | Computer Code · Code Generation | pass-1 | 85.0% | #5 | — | source ↗ |
| 04 | MMMU-Pro | Multimodal · Visual Question Answering | accuracy | 78.5% | #6 | — | source ↗ |
| 05 | SWE-Bench Verified | Computer Code · Code Generation | accuracy | 76.8% | #8 | — | source ↗ |
| 06 | OmniDocBench | Computer Vision · Document Parsing | accuracy | 88.8% | #9 | — | source ↗ |
| 07 | BrowseComp | Natural Language Processing · Question Answering | accuracy | 60.6% | #11 | — | source ↗ |
| 08 | GPQA Diamond | Reasoning · Multi-step Reasoning | accuracy | 87.6% | #12 | — | source ↗ |
| 09 | HLE | Reasoning · Multi-step Reasoning | accuracy | 30.1% | #17 | — | source ↗ |
| 10 | PLCC | Natural Language Processing · Polish Cultural Competency | grammar | 80.0% | #20 | — | source ↗ |
| 11 | PLCC | Natural Language Processing · Polish Cultural Competency | history | 89.0% | #22 | — | source ↗ |
| 12 | PLCC | Natural Language Processing · Polish Cultural Competency | geography | 86.0% | #34 | — | source ↗ |
| 13 | PLCC | Natural Language Processing · Polish Cultural Competency | average | 77.8% | #39 | — | source ↗ |
| 14 | PLCC | Natural Language Processing · Polish Cultural Competency | art-and-entertainment | 69.0% | #41 | — | source ↗ |
| 15 | PLCC | Natural Language Processing · Polish Cultural Competency | culture-and-tradition | 78.0% | #41 | — | source ↗ |
| 16 | PLCC | Natural Language Processing · Polish Cultural Competency | vocabulary | 65.0% | #53 | — | source ↗ |
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.
§ 03 · Strengths by area
Where Kimi-K2.5 actually performs.
§ 04 · Papers
1 paper with results for Kimi-K2.5.
- 2026-02-02· 9 results
Kimi K2.5: Visual Agentic Intelligence
§ 05 · Related models
Other Moonshot.AI models scored on Codesota.
§ 06 · Sources & freshness
Where these numbers come from.
pwc-dump
9
results
sdadas/PLCC
7
results
7 of 16 rows marked verified.