Kimi K2 Thinking
ModelOpen sourceActiveby Moonshot AI · family “kimi-k2-thinking”
Kimi K2 Thinking is Moonshot AI’s most advanced open reasoning model to date, extending the K2 series into agentic, long-horizon reasoning. Built on the trillion-parameter Mixture-of-Experts (MoE) architecture introduced in... (description from the OpenRouter listing)
in: textout: textReasoningTool useStructured output
Epoch Capabilities Index
146.0 #73 of 268
90% CI 143.0 – 147.4
Released
6 Nov 2025
Context window
262K tokens
Max output
98K tokens
Input $ / 1M tokens
$0.60
Output $ / 1M tokens
$2.50
Cached input $ / 1M
$0.15
Each benchmark shown separately with its own source
Benchmark results (4)
| Benchmark | Domain | Score | vs best recorded | Setting | Run | Source |
|---|---|---|---|---|---|---|
| CL-bench | long-context | 11.9% | 43% | — | — | External ↗ |
| Terminal Bench | agents | 35.7% ±2.8 | 42% | agent: Terminus 2 | 6 Nov 2025 | External ↗ |
| METR Time Horizons | agents | 59.2% | 69% | — | — | External ↗ |
| WeirdML | coding | 42.8% ±0.0 | 46% | — | — | External ↗ |
Source: Epoch AI Benchmarking Hub (CC BY 4.0). “External” rows are leaderboard results Epoch collects from third parties. Best reported setting per benchmark is shown.
API list price over time
Price history
$ per 1M tokens
11 recorded prices on 6 Dec 2025 (OpenRouter listing + Internet Archive snapshots); re-read on every data refresh, most recently 7h ago. Steps show when the price changed.
Events
- Open releaseMoonshotAI releases Kimi K2 ThinkingEpoch AI Benchmarking Hub