Kimi
Kimi K2
Kimi K2 is a state-of-the-art mixture-of-experts (MoE) language model with 32 billion activated parameters and 1 trillion total parameters. Trained with the Muon optimizer, Kimi K2 achieves exceptional performance across frontier knowledge, reasoning, and coding tasks while being meticulously optimized for agentic capabilities.
Kimi model releasesModel Snapshot
TypeLanguage
ReleaseJul 11, 2025
Context131,072
Tagstool-use
Overall
Rank #256 across 600
19.4
Coding
No rank
n/a
Blended Price
Rank #195 across 379
$1.00/M
Strength Profile
Percentile score by analysis domain.
* Cost is inverted: lower input, output, and blended prices rank higher.
Benchmark Percentiles
Higher bars mean stronger relative placement.
Model Details
Runtime
Max Output100,352
Streaming speed is not measured for this model yet.
Pricing
Blended$1.00/M
Input$0.570/M
Output$2.30/M
Catalog$0.570/M in / $2.30/M out
Metadata
Modalities
All Benchmarks
| Metric | Domain | Value | Rank |
|---|---|---|---|
| Artificial Analysis Intelligence Index | overall | 19.4 | #256 |
| Artificial Analysis Math Index | math | 57.0 | #124 |
| GPQA | reasoning | 76.6% | #198 |
| Humanity's Last Exam | reasoning | 7.0% | #276 |
| LiveCodeBench |