Moonshot · released Jan 27, 2026
Kimi K2.5
Kimi K2.5 is an older open Moonshot model that still handles long documents.
Open weights9 months old
- Input price
- $0.50 / M
- Output price
- $2.5 / M
- Context
- 262K tokens
- Longest reply
- –
- Reads
- text, image
- Knows up to
- –
Where it shines
- It scores 78.1 on a long-context test.
- It scores 63.3 on OSWorld computer use.
- It costs $0.50 in, $2.50 out per million tokens.
Where it struggles
- Its ARC-AGI-2 score of 11.8 is far behind.
- Its context tops out at 262,144 tokens.
Where you can use it
- Hugging FaceDownloadWeights are public.
- Moonshot APIAPI$0.50 in / $2.50 out per million tokens.
- OpenRouterAPIListed as
moonshotai/kimi-k2.5
How it scores on each job
Scores run 0 to 100 and blend every measure we have for that job. The number after the dot is its rank among the 33 models in the lineup.
The evidence
| Measure | Score | Rank in lineup | Source |
|---|---|---|---|
| Epoch Capabilities Index | 148 | #26 | Epoch AI |
| LMArena text | 1445 | #20 | LMArena |
| LMArena hard prompts | 1453 | #22 | LMArena |
| LMArena instruction following | 1431 | #23 | LMArena |
| LMArena multi-turn | 1444 | #23 | LMArena |
| LMArena creative writing | 1422 | #21 | LMArena |
| LMArena writing & language | 1425 | #23 | LMArena |
| LMArena coding | 1474 | #22 | LMArena |
| LMArena math | 1470 | #19 | LMArena |
| LMArena business & finance | 1427 | #23 | LMArena |
| LMArena legal & government | 1441 | #22 | LMArena |
| LMArena medicine | 1448 | #20 | LMArena |
| WebDev Arena | 1437 | #27 | LMArena |
| Vision Arena | 1269 | #17 | LMArena |
| Document Arena | 1430 | #9 | LMArena |
| SWE-bench Verified | 74% | #5 | Epoch AI |
| Terminal-Bench | 43% | #4 | Epoch AI |
| GPQA Diamond | 88% | #24 | Epoch AI |
| Humanity's Last Exam | 24% | #8 | Epoch AI |
| Fiction.LiveBench 120k | 78% | #1 | Epoch AI |
| SimpleQA Verified | 34% | #21 | Epoch AI |
| OSWorld | 63% | #1 | Epoch AI |
| ARC-AGI-2 | 12% | #21 | Epoch AI |
Compare it with these
New
Moonshot
Kimi K3.1 Pro
Kimi K3.1 Pro is Moonshot's newest model, and early tests put it near the top for coding and sites.
Websites & appsCoding
- Price
- $2 in · $10 out
- Context
- 1M
Announced
Gemini 4 Carbon
Gemini 4 Carbon isn't public yet, and Google reports it tops one independent intelligence index.
- Price
- –
- Context
- –
Open weights5 weeks old
Xiaomi
MiMo-V2.6-Pro
MiMo-V2.6-Pro is Xiaomi's open model and the strongest open pick for most jobs.
Everyday questionsWritingCoding
- Price
- $0.43 in · $0.87 out
- Context
- 1M