Gemini 4 Argon
Claude Opus 5.5
Muse Wonder
MiMo-V2.6-Pro
Qwen3.8-Max
GLM-5.3
GPT-5.5
ERNIE 5.1
DeepSeek-V4.1-Flash-20260910
Grok 4.20 Beta
amazon-nova-experimental-chat
Dola Seed 2.0 Pro
NVIDIA-Nemotron-3-Ultra-550B-A55B
Gemma 4 31B
Step 5 Preview
Kimi K2.5
Kimi K2.5 is an open-weight model from Moonshot AI, released Jan 27, 2026. It ranks #61 of 351 on the LMArena text leaderboard with a rating of 1445. It costs $0.50 per million input tokens and $2.50 per million output tokens, with a 256K-token context window.
- Developer
- Moonshot AI
- Released
- Jan 27, 2026
- Context window
- 256K
- Price in / out, per 1M tokens
- $0.50 / $2.50
- Weights
- Open · Modified MIT
- Inputs
- Text, Images
- Reasoning
- Yes
- Tool use
- Yes
- Hugging Face downloads, 30 days
- 313.6K
- Hugging Face likes
- 2.9K
Benchmarks
| Benchmark | Score bar | Accuracy | Rank | Cost / test |
|---|---|---|---|---|
| SWE-benchCoding | 70.00% | 58 / 85 | $0.17 | |
| Vibe Code BenchCoding | 17.54% | 88 / 115 | $0.88 | |
| LiveCodeBenchCoding | 83.87% | 46 / 143 | $0.6 / $3 | |
| Legal Research BenchLaw | 15.87% | 66 / 85 | $0.23 | |
| MedCodeMedicine | 39.32% | 69 / 107 | $0.6 / $3 | |
| MedQAMedicine | 94.37% | 17 / 92 | $0.6 / $3 | |
| GPQA DiamondReasoning | 84.09% | 55 / 138 | $0.6 / $3 | |
| MMLU-ProReasoning | 85.91% | 52 / 138 | $0.6 / $3 | |
| AIMEReasoning | 95.63% | 11 / 96 | $0.6 / $3 | |
| MMMUReasoning | 84.33% | 26 / 93 | $0.6 / $3 |
Vals AI, as of Oct 24, 2026.
LMArena, by category
| Leaderboard | Rating | Rank | Votes |
|---|---|---|---|
| TextLMArena | 1445 | 61 / 351 | 74.7K |
| CodingLMArena | 1484 | 50 / 346 | 1.9K |
| MathLMArena | 1470 | 46 / 339 | 4.2K |
| Creative writingLMArena | 1422 | 67 / 349 | 12.3K |
| Instruction followingLMArena | 1431 | 74 / 351 | 24.2K |
| Hard promptsLMArena | 1453 | 67 / 351 | 47.9K |
| Multi-turnLMArena | 1444 | 79 / 349 | 13.1K |
| Long queriesLMArena | 1444 | 72 / 330 | 29.8K |
| ExpertLMArena | 1466 | 64 / 305 | 7K |
| Legal and governmentIndustries | 1441 | 85 / 325 | 5.6K |
| Medicine and healthcareIndustries | 1448 | 77 / 325 | 5.2K |
| Business and financeIndustries | 1427 | 89 / 344 | 14.4K |
| ChineseLanguages | 1494 | 60 / 328 | 4.6K |
| SpanishLanguages | 1450 | 61 / 249 | 2.5K |
| GermanLanguages | 1441 | 70 / 261 | 1.3K |
| FrenchLanguages | 1454 | 79 / 252 | 2.6K |
| JapaneseLanguages | 1423 | 50 / 236 | 851 |
| WebDevLMArena | 1437 | 83 / 124 | 20.1K |
| VisionLMArena | 1269 | 46 / 146 | 28.2K |
LMArena, CC BY 4.0, as of Oct 22, 2026. Where a model is rated at several effort levels, its best is shown (Kimi K2.5 (Thinking), Kimi K2.5 (Instant)).
Compare Kimi K2.5
Moonshot AI
The lab behind Kimi, known for long-context and agentic open-weight models. Founded 2023, based in Beijing. moonshot.ai
Questions
Who makes Kimi K2.5?
Kimi K2.5 is made by Moonshot AI, founded 2023 and based in Beijing.
How much does Kimi K2.5 cost?
$0.50 per million input tokens and $2.50 per million output tokens, at its provider’s list price.
What is Kimi K2.5’s context window?
256K tokens (262,144).
How does Kimi K2.5 rank?
#61 of 351 on LMArena’s text leaderboard, rated 1445 (95% interval 1442–1448) from 74,744 votes, as of Oct 22, 2026.
Is Kimi K2.5 open source?
Kimi K2.5 has open weights under the Modified MIT license, downloadable from Hugging Face (moonshotai/Kimi-K2.5).