[docs] Add UomiRouter inference provider#2499
Closed
cuttini wants to merge 5 commits into
Closed
Conversation
added 5 commits
May 25, 2026 12:41
Companion to huggingface/huggingface.js#2193 and huggingface/huggingface_hub#4256. Adds the UomiRouter provider page, sidebar entry, partners-table row, and PROVIDERS_URLS constant. Logos are placeholder PNGs in this repo; real SVG/PNG light+dark assets will be uploaded to the documentation-images HF dataset before merge. Capabilities: Chat completion (LLM) ✅, Chat completion (VLM) ✅ (Gemma 4 VL behind the same /v1/chat/completions endpoint). Endpoint: https://gateway.uomi.ai (OpenAI Chat Completions spec; streaming, tool calling, structured output, vision).
- Rewrote provider description: GPUs currently run directly by UomiRouter
(not yet third-party operators); OPoC is off-chain today (signed
responses + sampled cross-operator re-verification); on-chain UOMI L1
anchoring is the next milestone, not live yet
- Renamed mapped models to HF canonical IDs (no -FP8 suffix, lowercase
per HF convention):
Qwen/Qwen3.6-27B-Instruct-FP8 -> Qwen/Qwen3.6-27B
google/gemma-4-31B-it -> google/gemma-4-31b-it
- Removed reference to vast.ai marketplace (currently we run the GPUs)
Clarify that inference is served by accredited operator nodes joining the UOMI network — each operator runs the engine on their own hardware after a hardware + reliability vetting. Privacy/obfuscation/OPoC commitments are part of the operator onboarding contract.
- google/gemma-4-31b-it -> google/gemma-4-31B-it (canonical, capital B; lowercase returns 307 redirect on huggingface.co) - providerModelId now reflects our INTERNAL model_id (qwen/qwen3.6-27b) rather than duplicating the HF canonical hfModel — matches how the mapping table is wired in Together/Fireworks etc.
Contributor
|
hi @cuttini, closing this based on this comment huggingface/huggingface.js#2193 (review) |
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
2 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
Add the UomiRouter provider page to the inference-providers docs, alongside the sidebar entry, partners table row, and PROVIDERS_URLS constant.
UomiRouter is an OpenAI-compatible distributed inference network where every response is signed by the GPU that produced it and anchored on-chain via UOMI L1's Proof of Computation (OPoC). Open-weight catalog at $0.10/Mtok.
Changes
scripts/inference-providers/templates/providers/uomirouter.handlebars— provider templatedocs/inference-providers/providers/uomirouter.md— rendered pagedocs/inference-providers/_toctree.yml— sidebar entry between Together and WaveSpeedAIdocs/inference-providers/index.md— partners-table row (Chat LLM ✅, VLM ✅)scripts/inference-providers/scripts/generate.ts—PROVIDERS_URLS["uomirouter"]entryassets/logos/uomirouter-{light,dark}.png— placeholders; real artwork to followCompanion PRs (required before this row renders)
generate.tsfiltersPROVIDERS_URLSbyPROVIDERS_HUB_ORGSimported from@huggingface/inference, so the JS PR must merge first for the row to be picked up.Logo note
Logos in
assets/logos/are ImageMagick-generated placeholders (brand color#dffe00). Real SVG + light/dark PNG assets will be uploaded to thedocumentation-imagesHF dataset and the page reference updated before merge.Reviewers
cc @Wauplin @SBrandeis @julien-c @hanouticelina per the new-provider checklist.
Note
Low Risk
Documentation and template-only changes with no runtime or security-sensitive code paths.
Overview
Adds UomiRouter to the Inference Providers docs: a new provider page (from
uomirouter.handlebars), sidebar entry in_toctree.yml, a partners-table row on the index (chat LLM and VLM), andPROVIDERS_URLS.uomirouteringenerate.tsso the doc generator can build the page.The page describes an OpenAI-compatible distributed network (gateway, OPoC signing headers, privacy/billing notes) and includes generated task snippets for chat LLM and VLM once live models exist in the Hub API.
Note:
generate.tsonly emits this provider afteruomirouterappears in@huggingface/inference’sPROVIDERS_HUB_ORGS(companion JS/hub PRs).Reviewed by Cursor Bugbot for commit c6f46da. Bugbot is set up for automated code reviews on this repo. Configure here.