API overview
ozeye mirrors the OpenAI API style while routing requests to supported EU providers. Every billable inference endpoint validates model, modality, provider configuration, pricing, balance, and key spend limits before routing.
Base URL
https://ozeye.ai/v1 for model, inference, account, notebook, and preset endpoints.
Auth
Use Authorization: Bearer $OZEYE_API_KEY. Dashboard sessions can also call account APIs from the web app.
Compatibility
Chat, embeddings, audio transcription, speech, and image generation use OpenAI-compatible request/response shapes where possible.
Zero data retention
Prompts and provider responses are never stored by ozeye. Usage metadata is kept for billing, see data retention.
API sections
- Chat completions - normal and streaming chat requests.
- Audio - transcription and speech synthesis.
- Embeddings - text-to-vector models.
- Image generation - supported image models.
- Web search and fetch - hosted search and page fetch endpoints.