Get up and running with Kimi, GLM, MiniMax, DeepSeek, gpt-oss, Qwen, Gemma and other models.
Commits on main
model: update lfm2 parser/renderer for optional thinking (#16359)
993acc7 Jeffrey Morgan committed 5 months ago
mlxrunner: harden linear/embedding layers against over-promotion (#16682)
82e0ddb Patrick Devine committed 5 months ago
mlxrunner: record committed MTP drafts before streaming them
1abd56b Jesse Gross committed 5 months ago
nn/recurrent: return per-boundary states from the gated-delta kernels
177aefb Jesse Gross committed 5 months ago
mlxrunner/cache: split KVCache and RotatingKVCache into their own files
07588c6 Jesse Gross committed 5 months ago
mlxthread: preserve the original stack when worker work panics
4c97a94 Jesse Gross committed 5 months ago
docs: update docs examples to use Gemma 4 instead of Gemma 3 (#16607)
f0078ae Jeffrey Morgan committed 5 months ago
Add AGENTS.md and CLAUDE.md to root repository (#16604)
96201a6 Jeffrey Morgan committed 5 months ago
launch: use native Windows Hermes config path (#16558)
a0099da Bruce MacDonald committed 5 months ago
docs: update Zod example to use native toJSONSchema (#14746)
25e0e81 Chris Chen committed 5 months ago
mlx: allow the embedding layer to use the nvfp4 global scale (#16527)
3ef69ef Patrick Devine committed 5 months ago
llama-server: allow GPU offload for projectors (#16473)
01557ff Daniel Hiltgen committed 5 months ago
mlx: "requires" in modelfile is being ignored for mlx based models (#16469)
e5a3873 Patrick Devine committed 5 months ago