The Gateway exposes multiple modalities through a unified API surface. Each endpoint maps to a different capability, and model support varies by provider.
Modalities and endpoints
| Modality | Primary endpoints | Notes |
|---|
| Text | /v1/responses, /v1/chat/completions, /v1/messages | Structured and conversational outputs. |
| Images | /v1/images/generations, /v1/images/edits | Text-to-image and image editing. |
| Audio | /v1/audio/speech, /v1/audio/transcriptions, /v1/audio/translations | Speech synthesis and speech-to-text. |
| Video | /v1/video/generation, /v1/video/status, /v1/video/content, /v1/video/delete | Generate and retrieve video outputs. |
| Music | /v1/music/generate | Music generation via supported providers. |
| OCR | /v1/ocr | Extract text from images where supported. |
Checking model support
Use the Models endpoint to see which models are available and which endpoints they support. Provider coverage is available via the Providers endpoint.
Best practices
- Match the endpoint to the modality you need, even if the model name is shared across modalities.
- Validate payloads against the API Reference before shipping.
Last modified on February 11, 2026
