You can now run Kimi K2.7 Code locally! đ
We shrank the 1T model to 325GB (-48%) via Dynamic 2-bit where important layers are upcasted.
Run at >40 tok/s on 330GB RAM/VRAM setups.
Run full precision on 610 GB.
Guide:
GGUF:
đ Kimi-K2.7-Code, our latest coding model, is now released and open-sourced!
đˇ Improved coding & agent performance over K2.6: +21.8% on Kimi Code Bench v2, +11.0% on Program Bench, and +31.5% on MLS Bench Lite.
đˇ Reasoning efficiency: Less overthinking, with 30% lower reasoning-token usage compared to K2.6.
đˇ Long-horizon coding: Improved instruction following, higher end-to-end coding task success rates.
âĄī¸ 6x High-Speed Mode coming soon!
đ Available today via Kimi API and Kimi Code.
đ Kimi Code:
đ API: