OpenAI-compatible providers
Many third-party services expose an OpenAI-compatible API (/v1/chat/completions, Bearer auth). Mystilink Agent can connect via a custom Provider or official template + Base URL override.
When to use which
| Approach | Best for |
|---|---|
| Official template + URL override | Proxy to the same vendor (e.g. OpenAI gateway) |
| Custom Provider | OpenRouter, SiliconFlow, Groq, Together, Azure OpenAI, … |
Set API Family to openai-compatible.
info
Anthropic and Gemini official APIs are not OpenAI-compatible — use their official templates.
Custom Provider steps
Settings → Models → New Provider
- Template: Custom provider
- Provider ID (e.g.
openrouter), display name, API Familyopenai-compatible - Default Base URL (include
/v1, e.g.https://openrouter.ai/api/v1) - API key → enable model discovery if supported → Create Provider
- Add model → test connection → model ID → save
Examples
| Service | Base URL |
|---|---|
| OpenRouter | https://openrouter.ai/api/v1 |
| SiliconFlow | https://api.siliconflow.cn/v1 |
| Groq | https://api.groq.com/openai/v1 |
| Together | https://api.together.xyz/v1 |
Azure OpenAI: use your resource deployment URL; model ID = deployment name.
Checklist
- Base URL path matches vendor docs (do not append
/chat/completions) - Test connection succeeds
- Model ID matches docs exactly