GPQA Diamond
?what it ishighest score on this metric 96.4%
38º / 198 · Epoch AI, latest version we checked
Benchmarks, price, context and versions
Kimi K2, kimi k2.6 pricing, kimi k2.6 size, kimi k2.6 context window
Direct answer: Kimi K2.6 is a model from Moonshot AI. 90.8% on GPQA Diamond, per Epoch AI. On OpenRouter, US$ 0.65 per 1 million input tokens and US$ 3.41 per 1 million output tokens.
This block follows a rule of this section, using source data: if the model is in the top 25% of the metric, the item goes under "Good for"; if it is in the bottom half, it goes under "Not the best for". The middle band appears in neither. Price and context use OpenRouter; scores use Epoch AI or OpenRouter Benchmarks, in the latest version we checked.
highest score on this metric 96.4%
38º / 198 · Epoch AI, latest version we checked
highest score on this metric 100%
38º / 186 · Epoch AI, latest version we checked
highest score on this metric 95.1%
46º / 86 · Epoch AI, latest version we checked
highest score on this metric 94%
56º / 87 · Epoch AI, latest version we checked
highest score on this metric 83.5%
9º / 32 · Epoch AI, latest version we checked
highest score on this metric 96.5%
44º / 135 · OpenRouter Benchmarks, latest version we checked
highest score on this metric 82%
54º / 131 · OpenRouter Benchmarks, latest version we checked
Dates come from the sources: OpenRouter listing or Hugging Face repo creation, which may differ from the official launch.
On OpenRouter, US$ 0.65 per 1 million input tokens and US$ 3.41 per 1 million output tokens.
262,144 tokens, per OpenRouter.
Yes, it has open weights. License: no data.
Kimi K2.6 is a model from Moonshot AI, with open weights, used for text, agents, code.
90.8% on GPQA Diamond, per Epoch AI.
It scores 76.7% on SWE-bench Verified, per Epoch AI.
It was made by Moonshot AI, a lab from China.
provisional data Questions ordered by what shows up in public Google search suggestions (no volume). Answers use source data only.
Dates come from the sources: OpenRouter listing or Hugging Face repo creation, which may differ from the official launch. Family and version were grouped by rule from the name.