Send images, documents, and rich media to models that support them.
The OpenDunes API supports multimodal inputs and outputs — image understanding, image and video generation, speech synthesis and transcription, audio and document inputs.
Image inputs — any model with the vision capability accepts images embedded in the messages[].content array, as a URL or base64 data alongside your text prompt. See Image Inputs.
| Feature | Page |
|---|---|
Image generation — POST /v1/images/generations | Image Generation |
| PDF document inputs | PDF Inputs |
| Audio inputs | Audio Inputs |
Video generation — POST /v1/videos | Video Generation |
Text-to-speech — POST /v1/audio/speech | Text-to-Speech |
Speech-to-text — POST /v1/audio/transcriptions | Speech-to-Text |
| Feature | Page |
|---|---|
| Video inputs | Video Inputs |
Use the capability filter on the model catalog to find models that support a given modality:
Browse the full catalog at /models. Each model page lists its supported input and output modalities so you can pick the right one before writing code.
Multimodal requests are billed the same way as text — per token, in Algerian Dinars (DA). Vision models typically count image tokens alongside text tokens; the exact token cost per image depends on the model and image resolution. The catalog shows DA-per-million-token rates for each model.
See Pricing for how DA credits work and how to top up your balance.