Loading...
Loading...
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 Kimi K2, it activates 32 billion parameters per forward pass and supports 256 k-token context windows. The model is optimized for persistent step-by-step thought, dynamic tool invocation, and complex reasoning workflows that span hundreds of turns. It interleaves step-by-step reasoning with tool use, enabling autonomous research, coding, and writing that can persist for hundreds of sequential actions without drift. It sets new open-source benchmarks on HLE, BrowseComp, SWE-Multilingual, and LiveCodeBench, while maintaining stable multi-agent behavior through 200–300 tool calls. Built on a large-scale MoE architecture with MuonClip optimization, it combines strong reasoning depth with high inference efficiency for demanding agentic and analytical tasks.
Quality Index
40.9
41st of 442
Top 9%
Coding Index
34.8
49th of 352
Top 14%
Math Index
94.7
10th of 268
Top 4%
Price/1M
$1.07
491st cheapest
247% above median
Top 72%
Speed
102 tok/s
Top 27%
TTFT
0.64s
Context Window
131K
145th largest
Top 63%
Input
$0.60
per 1M tokens
Output
$2.50
per 1M tokens
Blended
$1.07
per 1M tokens
Cheaper than 28% of models. Median price is $0.31/1M tokens.
Daily
$1.07
Monthly
$32.25
102
tokens/sec
Faster than 73% of models
0.64
seconds
Faster than 39% of models
20.24
seconds
Faster than 14% of models
Market Median
46 tok/s
124% faster
Median TTFT
0.42s
52% slower
Throughput/Dollar
95
tok/s per $/1M
Speed Comparison
Context Window
131K
tokens
Larger than 37% of models