Kimi K2 Thinking: standardized the context-window label to the documented 256K tokens
Moonshot AI aligns Kimi K2 documentation with verified 256K token capacity standards.
Standardized the context-window label to the documented 256K tokens — verified by the OnAirToday AI Progress Tracker against Moonshot's published documentation.
OnAirToday
Newsroom
The OnAirToday AI Progress Tracker recorded a verified change to Moonshot's Kimi K2 Thinking (version kimi-k2-thinking): Standardized the context-window label to the documented 256K tokens.
Kimi K2 Thinking is Moonshot's open-weight reasoning model for text-based problem solving, coding and multi-step tool use, with a documented 256K-token context window. The newer Kimi K2.5 belongs to the same family, but its specifications and results should not be attributed to K2 Thinking.
What changed in this review:
- Added vendor-published reasoning, coding and tool-assisted benchmark results with evaluation distinctions.
- Removed the unconfirmed Terminal-Bench 35.7% and FrontierMath 21.4% entries.
- Kimi K2.5 has shipped; this profile remains specific to Kimi K2 Thinking.
- Added separately attributed coverage of Kimi K3-assisted Redis exploit research.
- Pricing remains $0.60 per million uncached input tokens and $2.50 per million output tokens.
- Capability score 30 → 71
Teams building on Kimi K2 Thinking should confirm limits, pricing and benchmark coverage against the vendor's own documentation before relying on these figures.
The full profile — capabilities, benchmark citations, known flaws and risk guidance — is on the tracker at /ai-models/kimi-k2-thinking.
References:
- https://huggingface.co/moonshotai/Kimi-K2-Thinking
- https://github.com/MoonshotAI/Kimi-K2
- https://platform.moonshot.ai/docs/pricing/chat
- https://huggingface.co/moonshotai/Kimi-K2.5
Generated insight by OnAirToday AI
Standardizing context-window specifications is crucial for developers to build reliable applications without unexpected token limitations. This move by Moonshot brings much-needed transparency to the competitive landscape of high-capacity reasoning models.
Get the day's AI news in one short email. Free.
Comments
Keep comments relevant and respectful. No links, advertising, self-promotion, or spam.