Text Models
Chat Completions
POST /chat/completions matches the OpenAI Chat Completions spec. The following parameters are supported:
model, messages, max_tokens, temperature, top_p, stream, stop, frequency_penalty, presence_penalty, n, user, seed, logprobs, top_logprobs, response_format, tools, tool_choice
Other fields are silently ignored. max_tokens is capped at 8192; values higher than that are clamped. Request bodies larger than 500 KB are rejected with a 413.
Call GET /models (relative to the v1 base URL) with your API key to fetch the current list of available model IDs.
| Model ID | Context | Strengths |
|---|---|---|
| deepseek/deepseek-v4.1-flash | 1M | Fast general model: reasoning, tool calling, image input |
| deepseek/deepseek-r1-0528 | 128K | Deep chain-of-thought, math, analysis |
| qwen/qwen3.5-397b-a17b | 256K | Arabic, multilingual, long-doc, enterprise |
| deepseek/deepseek-v4-pro | 1M | Open-weight flagship: reasoning, agentic coding, STEM |