Quality
Highest overall scores
- 1
MoonshotAI: Kimi K357.1 - 2MoonshotAI: Kimi K2.7 Code41.9
- 3MoonshotAI: Kimi K2 Thinking17.3
Moonshot AI model benchmarks.
Compare Moonshot AI models across answer quality, coding, complex tasks, speed, price, and context size.
Response speed
Fastest output in this sample
- 1MoonshotAI: Kimi K2.7 Code168 t/s
- 2MoonshotAI: Kimi K2 Thinking79 t/s
- 3MoonshotAI: Kimi K373 t/s
Input price
Lowest price per 1M tokens
- 1MoonshotAI: Kimi K2 Thinking$0.60/1M
- 2MoonshotAI: Kimi K2.7 Code$0.73/1M
- 3MoonshotAI: Kimi K3$3.00/1M
Highlights
Quality
Overall capability · Higher is better
Speed
Output tokens per second · Higher is better
Input price
USD per 1M tokens · Lower is better
Quality benchmarks
Compare Moonshot AI model capability.
Performance and price
Compare practical tradeoffs.
All Moonshot AI models.
3 models in the current catalogue.
| Model | Links | |||||
|---|---|---|---|---|---|---|
| MoonshotAI: Kimi K3 | 57.1 | 73 t/s | 1M | $3.00 | $15.00 | |
| MoonshotAI: Kimi K2.7 Code | 41.9 | 168 t/s | 262.1K | $0.73 | $3.50 | |
| MoonshotAI: Kimi K2 Thinking | 17.3 | 79 t/s | 262.1K | $0.60 | $2.50 |
Moonshot AI technical benchmarks.
Compare named evaluation records available for Moonshot AI models. Missing results remain blank rather than being estimated.
| Benchmark | Area | Kimi K3 | Kimi K2.7 Code | Kimi K2 Thinking | Source |
|---|---|---|---|---|---|
| Intelligence Index | overallQuality | 57.1 | 41.9 | 17.3 | OpenRouter / artificial-analysis |
| Coding Index | coding | 76.2 | 60.8 | 21 | OpenRouter / artificial-analysis |
| Agentic Index | agentic | 50.1 | 29.6 | 1.8 | OpenRouter / artificial-analysis |
| GPQA | GPQA Diamond | 93.5 | 89.6 | 71.3 | OpenRouter model benchmarks |
| Humanity's Last Exam | HLE | 44.3 | 32.8 | 9.5 | OpenRouter model benchmarks |
| AA-LCR | AA-LCR | 74.7 | 66.3 | 52.7 | OpenRouter model benchmarks |
| GDPval-AA | GDPval-AA | 59.4 | 34.4 | 0 | OpenRouter model benchmarks |
| CritPt | CritPt | 23.4 | 10 | 0 | OpenRouter model benchmarks |
| SciCode | SciCode | 58.7 | 47.5 | 33 | OpenRouter model benchmarks |
| AA-Omniscience Accuracy | AA-Omniscience Accuracy | 46 | 38.6 | 15.7 | OpenRouter model benchmarks |
| AA-Omniscience Non-Hallucination Rate | AA-Omniscience Non-Hallucination Rate | 49.1 | 19.7 | 41.1 | OpenRouter model benchmarks |
| IFBench | IFBench | — | 63.1 | 62.8 | OpenRouter model benchmarks |
| τ²-Bench Telecom | τ²-Bench Telecom | — | 90.1 | 25.4 | OpenRouter model benchmarks |
| Terminal-Bench Hard | Terminal-Bench Hard | — | 44.7 | 6.8 | OpenRouter model benchmarks |
| MMLU-Pro | Knowledge and reasoning | — | — | 84.6 | OpenEvals/leaderboard-data |
| SWE-bench Verified | Coding | — | — | 71.3 | OpenEvals/leaderboard-data |
| Terminal-Bench | Agentic coding | — | — | 35.7 | OpenEvals/leaderboard-data |