Kimi K2 Thinking
Thinking Kimi model for slower research passes, planning, and hard technical questions
Shogo Agent Index
Estimated#70 of 9230.3range 19.8–40.9 · 23% of weight measured
Agentic coding
32
23% measured
Long-horizon autonomy
37
50% measured
Knowledge-work agents
26
estimated
Computer use
20
estimated
Security & context learning
33
50% measured
Agentic benchmarks
| Benchmark | Score | Normalised | Index weight | Harness / source |
|---|---|---|---|---|
| SWE-bench Verified | est. | 8% | predicted 66 | |
| Terminal-Bench 2.0 | 35.7% | 8% | Terminus 2 | |
| DeepSWE | est. | 6% | predicted 25 | |
| FrontierSWE | est. | 5% | predicted 4 | |
| CursorBench | est. | 4% | predicted 27 | |
| GSO-Bench | est. | 4% | predicted 9 | |
| METR Time Horizons | 59.2% | 10% | source | |
| Vending-Bench 2 | est. | 6% | predicted 13 | |
| PostTrainBench | est. | 4% | predicted 17 | |
| APEX-Agents | est. | 8% | predicted 25 | |
| GDPval | est. | 7% | predicted 31 | |
| Remote Labor Index | est. | 5% | predicted 1 | |
| DeepResearch Bench | est. | 5% | predicted 45 | |
| OSWorld 2.0 | est. | 6% | predicted 3 | |
| OSWorld | est. | 4% | predicted 45 | |
| Cybench | est. | 5% | predicted 48 | |
| CL-bench | 17.6% | 5% | source |
Other benchmarks
| Benchmark | Category | Score | Source |
|---|---|---|---|
| GPQA Diamond | Reasoning | 84.2% | source |
| OTIS Mock AIME | Math | 83.1% | source |
| SimpleBench | Reasoning | 39.6% | source |
| WebDev Arena | Coding | 1337 | source |
API hosts
Live prices and uptime for every host serving Kimi K2 Thinking via OpenRouter.
| Host | Input | Output | Cache read | Agent $/M | Uptime 24h | Quant |
|---|---|---|---|---|---|---|
| Novita | $0.60 | $2.50 | $0.15 | $0.33 | 100.0% | bf16 |