ModelsIndexCompareBest forBenchmarksStatusPricingAPI

Kimi K2 Thinking

Moonshot AI·Operational·Released 2025-11-06text
Context256k
Max output-
Input / 1M-
Output / 1M-
Cached in-
Benchmarks17

Benchmarks

Cited public sources · rank is exact-benchmark, apples-to-apples

WeirdML (Håvard Tveit Ihle)#62/9242.8% ↗
SimpleBench (SimpleBench)#60/7339.6% ↗
AIME25 (no tools)94.5% ↗
BrowseComp (w/ tools)60.2% ↗
BrowseComp-ZH (w/ tools)62.3% ↗
GPQA (no tools)#2/484.5% ↗
HLE (Text-only) (no tools)23.9% ↗
HLE (Text-only) (w/ tools)44.9% ↗
HMMT25 (no tools)89.4% ↗
IMO-AnswerBench (no tools)78.6% ↗
LiveCodeBenchV6 (no tools)83.1% ↗
MMLU-Pro (no tools)84.6% ↗
Multi-SWE-bench (w/ tools)41.9% ↗
SciCode (no tools)44.8% ↗
SWE-bench Multilingual (w/ tools)61.1% ↗
SWE-bench Verified (w/ tools)71.3% ↗
Terminal-Bench (w/ simulated tools (JSON))47.1% ↗

Compare

vs Gemini 4 Argonvs GPT-6.1 Solvs Claude Sonnet 5.5vs Muse Spark 1.3vs GPT-6 Solvs GPT-6 Lunavs Claude Opus 5.5vs GPT-6 Astra

More from Moonshot AI

Kimi K3Kimi K2.7 CodeKimi K2.6Kimi K2.5Kimi K2-0905Kimi K2

Kimi K2 Thinking FAQ

What is Kimi K2 Thinking?
Kimi K2 Thinking is a large language model from Moonshot AI, released on November 6, 2025. It accepts text input.
When was Kimi K2 Thinking released?
Moonshot AI released Kimi K2 Thinking on November 6, 2025.
What is Kimi K2 Thinking's context window?
Kimi K2 Thinking has a 256K-token context window.
How does Kimi K2 Thinking score on benchmarks?
Kimi K2 Thinking has 17 cited benchmark scores, including WeirdML (Håvard Tveit Ihle) 42.8%, SimpleBench (SimpleBench) 39.6% and AIME25 (no tools) 94.5%. Each score links to its source.

Access Kimi K2 Thinking and every other model through one endpoint with automatic failover: Respan gateway.