Kimi K2 Thinking
by Moonshot AI
+
+
+
+
About
Kimi K2 Thinking, released by Moonshot AI in November 2025, adds extended chain-of-thought reasoning to the Kimi K2 architecture — designed for problems that benefit from deliberate, step-by-step reasoning rather than fast generation. It targets use cases where K2's agentic capabilities need to be paired with deeper reasoning over complex technical or analytical tasks.
+
+
+
+
Pricing Range
Input (per 1M)$1.00 -$1.00
Output (per 1M)$4.00 -$4.00
Providers1
+
+
+
+
Timeline
ReleasedNov 6, 2025
+
+
+
+
License & Family
License
Modified MIT
Performance Overview
Performance metrics and category breakdown
Overall Performance
1 benchmarks
Average Score
43.8%
Best Score
43.8%
High Performers (80%+)
0Performance Metrics
Max Context Window
264.2KTop Categories
Coding
43.8%
+
+
+
+
All Benchmark Results for Kimi K2 Thinking
Complete list of benchmark scores with detailed information
| SWE-rebench | Coding | 43.80 | 43.8% | Unverified |