Kimi K2 Thinking
Moonshot AI🇨🇳 China
Moonshot AI's open-weights reasoning model in the Kimi K2 series, built for agentic, long-horizon tasks.
Context window262K tokens
Input / 1M tokens$0.6
Output / 1M tokens$2.5
Pricing and context window via OpenRouter, 2026-11-20.
Version History
launchmajor
Kimi K2 Thinking released. Moonshot AI's open-weights reasoning model in the Kimi K2 series, built for agentic, long-horizon tasks.
Benchmark Scores
Full leaderboard →| Benchmark | Score | Rank | Source |
|---|---|---|---|
| GPQA | 83.8% | 96 of 165 | Artificial Analysis, Jul 2026 ↗ |
| MMLU-Pro | 84.8% | 29 of 76 | Artificial Analysis, Jul 2026 ↗ |
| LiveCodeBench | 85.3% | 10 of 48 | Artificial Analysis, Jul 2026 ↗ |
Rank is against every model in our database with a sourced score on that benchmark, each counted once with its best score.