Kimi K3 Tech Blog: Open Frontier Intelligence
Kimi released Kimi K3, a 2.8T-parameter open model with native vision and 1M-token context, available today on Kimi.com, Kimi Work, Kimi Code, and the Kimi API.
- Kimi K3 activates 16 of 896 experts via Stable LatentMoE, with KDA and Attention Residuals architecture.
- API pricing is $0.30/MTok cache-hit input, $3.00/MTok cache-miss input, and $15.00/MTok output.
- Full model weights release by July 27, 2026; technical report to follow.
- Kimi K3 designed a chip in a 48-hour autonomous run, closing timing at 100 MHz within 4 mm².
- Kimi K3 trails Claude Fable 5 and GPT 5.6 Sol overall but outperforms other tested models.
Open-source developers and researchers now have access to a frontier-scale 2.8T-parameter model with native vision and long context, shifting the open-model capability ceiling.
Sources
Read this as text
Back to the AI news