Skip to content

feat(gallery): add Qwen3.8 NVFP4 MTP variant - #11606

Open
localai-org-maint-bot wants to merge 1 commit into
masterfrom
cron/model-gallery-20260819-run4
Open

feat(gallery): add Qwen3.8 NVFP4 MTP variant#11606
localai-org-maint-bot wants to merge 1 commit into
masterfrom
cron/model-gallery-20260819-run4

Conversation

@localai-org-maint-bot

Copy link
Copy Markdown
Collaborator

Description

Add the Qwen3.8 27B NVFP4 medium-tier GGUF as a variant of the existing Qwen3.8 gallery entry. The model contains an embedded MTP head, enables draft-mtp, and includes its matching BF16 vision projector.

The model and projector SHA-256 values were verified against Hugging Face x-linked-etag headers.

Notes for Reviewers

  • The mtp tag matches the configured spec_type:draft-mtp option.
  • The BF16 projector uses a distinct local path from the existing Q8_0 projector.
  • Verification: go test ./core/gallery -ginkgo.focus="gallery/index.yaml" -count=1
  • The full gallery suite reached 375 passing specs; three unrelated network-fixture specs failed because this host blocks raw GitHub and gist URLs.

Signed commits

  • Yes, I signed my commits.
  • Documentation updated (docs/content/) for user-facing changes, or not applicable

Add a compact NVFP4 build with its embedded MTP head and vision
projector. This gives Blackwell-capable llama.cpp hosts a speculative
option alongside the existing Q4 and Q8 builds.

Assisted-by: Codex:gpt-5
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant