Text Models
Text models use POST /v1/chat/completions and return a response synchronously or as an SSE stream. They share the same HiAPI API key as media models, but do not use the asynchronous /v1/tasks request shape.
| Model | Best for | Endpoint |
|---|---|---|
| DeepSeek V4 Flash | High-throughput generation, reasoning, coding, and Agent workflows | POST /v1/chat/completions |
Start with the Chat Completions guide or compare the AI Text APIs collection.