Kimi K3
Kimi K3
The largest open-weights model anyone has shipped, and it reads a million tokens without blinking.
Updated
The read
Moonshot published K3 at a scale nobody else has open-sourced, with native vision, an explicit thinking mode and a million-token window. On long-document work and agentic tasks it is genuinely near the top of the field. Two caveats matter: the licence restricts commercial use above a revenue threshold, so it is open weights rather than open source, and the hosted API is priced like a frontier model rather than a Chinese one.
Where it shines
- Million-token context handled properly rather than nominally
- Native vision and an explicit thinking mode
- The strongest open-weights model on agentic benchmarks
Worth knowing first
- The licence caps commercial use above a revenue threshold
- Hosted API is priced well above other Chinese frontier models
Index score
87.5
Provisional. This score is read from the product's public capability, not from the one-prompt rebuild the fully reviewed entries went through, so treat it as a placing rather than a verdict.
- Rank
- 05 of 53
- Builds
- LLMs
- Category
- Open weights
- Output
- Published weights with licence conditions, plus an API
- Pricing
- Free tier
- Best for
- Long-document and agent work at the top of the open field
Scores are our own editorial judgement, weighted craft 55, speed 10, control 15 and value 20.
Compare it with
Also worth a look
Full index- 06GLM-5.2GLM-5.2ProvisionalFrontier-class output, an MIT licence, and a million tokens of context. That combination is the story.Open weightsFree tier87.1
- 12Qwen 3.6Qwen 3.6ProvisionalThe open Qwen release most people should actually run, in sizes that fit real hardware.Open weightsFree tier83.3
- 14Mistral Large 3Mistral Large 3ProvisionalEurope's answer, published under Apache 2.0, with a quarter of a million tokens of context.Open weightsFree tier82.9
