King's e-Research AI Hub API

An OpenAI-compatible LLM inference API operated by King's e-Research for researchers, students and staff. Per King's documentation it serves chat completions against a curated set of hosted models, image generation, streaming responses, RAG pipelines over user documents, deep-research and agent workflows, and integrations with VS Code, OpenCode and MCP servers — all reachable by pointing an official OpenAI client at a new base URL with an AI Hub API key. New accounts receive an allocation of free tokens; sustained use is metered against a project budget. Verified live from outside: GET /api/v1/models answers 401 application/json with the OpenAI error envelope, and answers differently for a missing header than for an unknown key. No OpenAPI is saved for this surface, because King's publishes none — the API reference at /docs is behind Microsoft Entra ID sign-in, and authoring a contract from two probed responses would credit King's with a document it has not written. This is the single most consequential finding of the re-profile: a research university that built and operates its own agent-facing inference API.

API entry from apis.yml

apis.yml Raw ↑
aid: kings-college-london:er-ai-hub
name: King's e-Research AI Hub API
x-operator: institution
x-operator-evidence: ai.create.kcl.ac.uk is a CNAME to lbext-vip.er.kcl.ac.uk -> 193.61.202.139, King's
  own e-Research load balancer, the same origin that serves docs.er.kcl.ac.uk, er.kcl.ac.uk and portal.er.kcl.ac.uk.
  No vendor CNAME, no shared platform host. The service is described in King's own documentation as "King's
  College London's AI inference platform for researchers", support runs to support@er.kcl.ac.uk, and accounts
  are billed against King's project budgets. This is King's engineering on King's infrastructure.
description: 'An OpenAI-compatible LLM inference API operated by King''s e-Research for researchers, students
  and staff. Per King''s documentation it serves chat completions against a curated set of hosted models,
  image generation, streaming responses, RAG pipelines over user documents, deep-research and agent workflows,
  and integrations with VS Code, OpenCode and MCP servers — all reachable by pointing an official OpenAI
  client at a new base URL with an AI Hub API key. New accounts receive an allocation of free tokens;
  sustained use is metered against a project budget. Verified live from outside: GET /api/v1/models answers
  401 application/json with the OpenAI error envelope, and answers differently for a missing header than
  for an unknown key. No OpenAPI is saved for this surface, because King''s publishes none — the API reference
  at /docs is behind Microsoft Entra ID sign-in, and authoring a contract from two probed responses would
  credit King''s with a document it has not written. This is the single most consequential finding of
  the re-profile: a research university that built and operates its own agent-facing inference API.'
humanURL: https://docs.er.kcl.ac.uk/CREATE/ai_hub/
baseURL: https://ai.create.kcl.ac.uk/api/v1
tags:
- Artificial Intelligence
- LLM
- Inference
- Research Computing
- OpenAI-Compatible
tags_raw:
- Artificial Intelligence
- LLM
- Inference
- Research Computing
- OpenAI Compatible
properties:
- type: Documentation
  url: https://docs.er.kcl.ac.uk/CREATE/ai_hub/
- type: Website
  url: https://ai.create.kcl.ac.uk/
- type: APIReference
  url: https://ai.create.kcl.ac.uk/docs
- type: Authentication
  url: authentication/kings-college-london-authentication.yml
- type: Errors
  url: errors/kings-college-london-problem-types.yml
- type: Support
  url: https://docs.er.kcl.ac.uk/CREATE/getting_help/