feat: register VibeThinker 1.5B / 3B (transformers + vLLM)#5085
Merged
Conversation
Register WeiboAI VibeThinker 1.5B and 3B in llm_family.json under both Transformers and vLLM engines with pytorch / none quantization. The models are Qwen2-architecture reasoning checkpoints; metadata (context length, chat template, model_revision) is pulled from the official HF / ModelScope repos. Also add .inference_home/ to .gitignore so that the local XINFERENCE_HOME cache used during development does not get accidentally committed.
![gemini-code-assist[bot]](https://avatars.githubusercontent.com/in/956858?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Code Review
This pull request updates the .gitignore file to exclude the .inference_home directory from version control. There are no review comments, and I have no feedback to provide as the change is correct and straightforward.
Important
The consumer version of Gemini Code Assist on GitHub is being sunset. Starting June 18, 2026, new organization installations will be blocked, and all code review activity will officially cease on July 17, 2026.
For more details on the timeline and next steps, please review the Help Documentation.
xiaoyesoso
pushed a commit
to xiaoyesoso/inference
that referenced
this pull request
Jun 28, 2026
Resolve conflict in xinference/model/llm/llm_family.json: both branches added a new model entry at the end of the array. Keep both entries: - Nex-N2 (from feat/nex-n2) - vibethinker (from upstream/main xorbitsai#5085) Also pulls in upstream changes: VibeThinker registration (xorbitsai#5085), models JSON update (xorbitsai#5093), stale running models fix (xorbitsai#5087), metrics label unification (xorbitsai#5075).
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
What
Register WeiboAI/VibeThinker 1.5B and 3B in
llm_family.jsonunder both Transformers and vLLM engines withpytorch / nonequantization. Metadata (architecture, context length, chat template,model_revision) is taken from the official HF / ModelScope repos.Also add
.inference_home/to.gitignoreso the localXINFERENCE_HOMEcache used during development is not committed.Why
VibeThinker is a small Qwen2-architecture reasoning checkpoint that runs fine on a single consumer GPU; adding it to the built-in registry lets users launch it with one click via the Web UI / CLI without crafting a custom registration file.
Notes
This PR is split out of #5084 to keep the change scoped to model registration. The two unrelated runtime fixes from that PR are sent separately:
DynamicCachehandling: seefix/transformers-dynamic-cachelist_modelsstale-replica guard: seefix/supervisor-stale-replicaTest
Registered both sizes locally, launched with Transformers and vLLM engines, verified chat completion works.