Skip to content

vMLX 1.6.48

Choose a tag to compare

View all tags
@jjang-ai jjang-ai released this 30 Aug 22:41
v1.6.48
cf23f7e

vMLX 1.6.48

Changes

  • Publish vMLX 1.6.47 updater manifest (4fbd20af)
  • Prototype combined GLM KDA decode fusion (5e831ccc)
  • Revert "Prototype combined GLM KDA decode fusion" (5bb7248a)
  • Prototype GLM native MTP prompt priming (bc4e0f2d)
  • Revert "Prototype GLM native MTP prompt priming" (8b20895e)
  • Prototype vectorized GLM KDA MTP verification (f4249c98)
  • Enable vectorized GLM KDA MTP verification (f4eaa071)
  • Make native MTP restart cache proofs configurable (fc88a3ea)
  • Add typed prompt SSD cache to MTP benchmark (d36bb5fd)
  • Attest fast gateway replays from durable request IDs (ee93ada9)
  • Keep Hugging Face download progress refresh-compatible (83deae56)
  • Prove fixed native MTP depth through the live UI (7d45992e)
  • Prepare vMLX 1.6.48 (057c3db4)
  • Retry transient release verification cleanup (cf23f7eb)

Runtime provenance

  • vMLX source: cf23f7eb2bee76037653c0e2aab7800d79b7e18f
  • JANG runtime: 54b29d29916fefa60ce15dce72e3735757f996eb (jang 2.5.47)
  • Tahoe/macOS 26 is the default build; Sequoia is the compatibility build.
  • Both DMGs are Developer ID signed, independently notarized, stapled, and Gatekeeper checked by the candidate workflow.

Downloads

  • Tahoe: vMLX-1.6.48-tahoe-arm64.dmg
    • SHA-256: 4fed26ff8f1989518abd876a930c07ef20f7124ef848258e541667c83b72ba2b
    • Size: 537978932 bytes
  • Sequoia: vMLX-1.6.48-sequoia-arm64.dmg
    • SHA-256: ab6286d7f9b5d66ae29619f5bebec7a46a0bfe1058aa33d0d035e832ce900f79
    • Size: 515902096 bytes

Python distributions

  • Wheel: vmlx-1.6.48-py3-none-any.whl — SHA-256 9e83d8310bd0bbea7faf27c6a5d26bf540bfd86c84e9a058769746de2058e58b
  • Source: vmlx-1.6.48.tar.gz — SHA-256 29503d03808f5179facc76a249afb11b2c9daf761bdeb67f20c1edebd7acdd99
Assets 9
Loading