Skip to content

vMLX 1.6.15

Choose a tag to compare

View all tags
@jjang-ai jjang-ai released this 22 Jul 09:11
v1.6.15
95135ba

vMLX 1.6.15 is a signed reliability and runtime checkpoint for Apple silicon.

Highlights:

  • Separate reasoning, visible-content, and tool-call streaming across Electron, OpenAI Chat/Responses, Anthropic, and Ollama paths.
  • Hardened tool continuation, parser buffering, session defaults, and single-model gateway swaps.
  • Laguna S2.1 q4 TurboQuant mixed-SWA cache restore, including disk-only partial-prefix reuse.
  • Targeted cache/runtime improvements for LFM, MiniMax M3, Nemotron Omni, Gemma 4 media, Qwen MTP video, and DSV4 long prefill.

Both Sequoia and Tahoe artifacts are Developer ID signed, Apple-notarized, stapled, Gatekeeper-accepted, and installed-smoked through the real Electron Start/Stop controls.

This is a usable public checkpoint; retained broad family/media/gateway stress rows remain documented follow-up work.

Thanks to @hornsan1 for continued testing and feedback.

Contributors

hornsan1
Assets 6
Loading