Skip to content

docs: add AbliteratedLLM endpoint guide - #937

Open
whiteworthy wants to merge 2 commits into
LibreChat-AI:mainfrom
whiteworthy:docs/abliteratedllm-endpoint
Open

whiteworthy wants to merge 2 commits into
LibreChat-AI:mainfrom
whiteworthy:docs/abliteratedllm-endpoint

Conversation

@whiteworthy

Copy link
Copy Markdown

Disclosure: I run AbliteratedLLM.

Closes #936

Summary

Adds an AbliteratedLLM custom endpoint guide under the LibreChat YAML AI endpoints docs, following the format of the NEAR AI Cloud page (#668).

The example uses:

  • ABLITERATEDLLM_API_KEY
  • https://api.abliteratedllm.dev/v1
  • glm-5.3-flash-abliterated as the default model and titleModel
  • models.fetch: true, since /v1/models is public and returns the live list

Every key in the example (name, apiKey, baseURL, models.default, models.fetch, titleConvo, titleModel, modelDisplayLabel) and the keys named in the notes (titleTiming, apiKey: "user_provided") are documented in the custom endpoint object structure page. There is no dropParams or addParams. LibreChat's stop-sequence control already caps at 4 entries, which is the API's limit. For these model IDs, LibreChat sends the output limit as max_tokens only and reasoning as reasoning_effort, and the API accepts both.

The Notes section covers three things a LibreChat user could hit:

  • The reasoning-effort control offers none and minimal. glm-5.3-flash-abliterated rejects both, and the other GLM models reject none.
  • Trial accounts allow one concurrent request, so the default titleTiming: "immediate" title call can get a 429. The note suggests titleTiming: "final" or titleConvo: false.
  • Image input only works on models that list image in input_modalities.

Other changes:

  • meta.json: abliteratedllm added in alphabetical order. The 13 meta.<locale>.json sidebars were regenerated with node scripts/sync-locale-meta.mjs --write so pnpm check:locale-meta stays green. Each gets the same one-line addition.
  • index.mdx: card added to "Other compatible endpoints" (AbliteratedLLM isn't in KnownEndpoints).
  • Second commit: brand icon (components/icons/providers/abliteratedllm.tsx, export, lib/icons.tsx map entry, and icon: frontmatter), in the same form as the NEAR icon. It's a separate commit so it can be dropped without touching the page.

This PR was prepared with an AI coding agent on my behalf.

Testing

  • node scripts/sync-locale-meta.mjs --write (regenerated the translated sidebars), then --check: "translated sidebars match the English pages"
  • npx prettier@3 --check on the changed .tsx, .ts and meta.json files, using the repo's .prettierrc.js: clean
  • git diff --check origin/main...HEAD: clean
  • curl -fsSL https://api.abliteratedllm.dev/v1/models | jq -e '.data | map(.id) | contains(["glm-5.3-flash-abliterated"])' → true

Not run: pnpm typecheck, pnpm lint and pnpm build, which need the full install. The librechat.yaml entry has not been tested in a running LibreChat instance, and no API key was used. The details in the notes come from the provider's API behaviour, not from a LibreChat run.

@vercel

vercel Bot commented Nov 14, 2026

Copy link
Copy Markdown

@whiteworthy is attempting to deploy a commit to the LibreChat's projects Team on Vercel.

A member of the Team first needs to authorize it.

This branch has not been deployed

No deployments
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.

docs: add an AbliteratedLLM custom endpoint example

1 participant