API Calls

This page covers all model capabilities of the platform, mapping one-to-one to the categories in Models. Start with the endpoint quick reference, then dive into the calling details of each capability as needed.

Endpoint Quick Reference

CapabilityMethodEndpointNotes
Text chatPOST/v1/chat/completionsChat, multi-turn, streaming
Vision understanding (multimodal)POST/v1/chat/completionsImages / video in messages; same endpoint as text chat
Text embeddingsPOST/v1/embeddingsText vectorization (text-embedding-v3 / v4)
RerankPOST/v1/rerankRe-ranks retrieval results (qwen3-rerank)
Text-to-speech (TTS)POST/v1/audio/speechText to speech (volcengine-tts-v1 / v2)
Music generationPOST/volcengine/music/submitAsync task: songs / background music (doubao-music-song / doubao-music-bgm), submit → poll → fetch result
Speech-to-text (STT)POST/v1/audio/transcriptionsSpeech recognition (audio uploaded as multipart)
Image generationPOST/v1/images/generationsText-to-image / image-to-image / Seedream layer decomposition (qwen-image-2.0, doubao-seedream-4-5, etc.)
Image generation (Vega async)POST/v1/tasks/vegaVega-series async task: submit → poll → download (wand-vega-image-*)
Image editingPOST/v1/images/editsUpload image + edit description (qwen-image-edit-*)
Video generationPOST/v1/video/generationsAsync task: submit → poll → fetch result (MiniMax-H3 also reuses this endpoint)
Video generation (Kling kling series)POST/v1/videosUnified async tasks for kling-* / kling-v-* models: submit → poll & fetch on GET /kling-video/v1/videos/{task_id}

For available models and real-time prices of each endpoint, see the Model Plaza; for model categories, billing methods, and naming conventions, see Models; for authentication, see Base URL & Authentication.

Capability Pages