docs(models): document Eden AI as an EU OpenAI-compatible gateway

[Victor M. SMITH (MVS-source)]

What

Documents Eden AI as a hosted OpenAI-compatible gateway in the Models & providers guide. No code change is needed: OpenScience already loads any OpenAI-compatible provider from a config.provider block (same mechanism as the local-models guide), so this adds a worked example plus the EU data-residency angle.

Eden AI is a French, EU-based OpenAI-compatible gateway. Example openscience.json:

{
  "provider": {
    "edenai": {
      "name": "Eden AI",
      "npm": "@ai-sdk/openai-compatible",
      "options": { "baseURL": "https://api.edenai.run/v3", "apiKey": "your-edenai-key" },
      "models": {
        "openai/gpt-4o-mini": {},
        "anthropic/claude-sonnet-4-5": {},
        "mistral/mistral-small-latest": {}
      }
    }
  }
}

Model ids are vendor-prefixed, so they run as edenai/openai/gpt-4o-mini, edenai/anthropic/claude-sonnet-4-5, etc.

Why Eden AI (EU / GDPR / data residency)

• EU data residency: a dedicated EU endpoint (https://api.eu.edenai.run/v3) that processes and routes prompts and outputs within the European Union (non-EU models are rejected).
• Zero data retention: prompts and outputs are not stored by default and are removed within 24 hours.
• Compliance: SOC 2 and ISO 27001, DPA as standard, designed around GDPR and the EU AI Act.

Refs: https://www.edenai.co/post/eu-ai-endpoint-how-to-keep-ai-requests-and-data-in-europe and https://www.edenai.co/data-compliancy

Changes

• frontend/docs/src/content/openscience/models.mdx: a "Hosted OpenAI-compatible gateways (EU data residency)" section with the Eden AI config block, mirroring the existing local-models config style.

Testing

• The documented mechanism is verified: OpenScience loads OpenAI-compatible providers through @ai-sdk/openai-compatible with the block's baseURL. I confirmed against the real Eden AI API with a real key using that exact adapter (createOpenAICompatible({ baseURL: "https://api.edenai.run/v3", apiKey })), which returns completions for openai/gpt-4o-mini, anthropic/claude-sonnet-4-5, and mistral/mistral-small-latest (the vendor-prefixed id is sent verbatim, so no double prefix is needed here).

I saw CONTRIBUTING asks to open an issue/discussion before larger changes; happy to convert this to an issue first, or to adjust the wording/placement.

[vercel]

Victor M. SMITH (@MVS-source) is attempting to deploy a commit to the InkVell Team on Vercel.

A member of the Team first needs to authorize it.