DeepSeek
DeepSeek · 1 model · Text → Text
https://onerouter.com/v1/chat/completionsHow to call
Pass any model ID from this series in the model field, e.g. deepseek/deepseek-v4-pro. The example on the right uses DeepSeek V4 Pro.
Overview
DeepSeek provides strong reasoning, tool use, and structured outputs with long-context support.
This series includes 1 model: deepseek/deepseek-v4-pro. They share the same request shape — switching between them is a one-line change to the model field of the Chat Completions API.
Models in this series
Switch tabs to compare each model's ID, context window, inputs, outputs, and capabilities. Check the Models page for current availability and pricing before use.
deepseek/deepseek-v4-pro| Provider | DeepSeek |
| Context window | 1.05M |
| Max output | 384K |
| Input | Text |
| Output | Text |
| Summary | DeepSeek V4 Pro delivers strong reasoning, tool use, and structured outputs with a 1.05M-token context. |
| Strengths | Strong reasoning with a very large maximum output for long-form generation. |
| Supported parameters | `frequency_penalty`, `logit_bias`, `logprobs`, `max_tokens`, `min_p`, `presence_penalty`, `reasoning` |
Request headers
| Parameter | Type | Required | Description |
|---|---|---|---|
| Authorization | string · required | required | Bearer token, formatted as `Bearer YOUR_API_KEY`. |
| Content-Type | string · required | required | Must be `application/json`. |
Request body
These models use the OpenAI-compatible Chat Completions schema. Full field reference:
| Parameter | Type | Required | Description | Values |
|---|---|---|---|---|
| model | string | required | Model ID from this series (see the tabs above). | |
| messages | array | required | Conversation so far. Each item has `role` (`system`/`user`/`assistant`/`tool`) and `content`. | |
| max_tokens | integer | optional | Maximum tokens to generate. Newer models use `max_completion_tokens`. | |
| temperature | number | optional | Sampling temperature, typically 0–2. Higher is more random. | |
| top_p | number | optional | Nucleus sampling, 0–1. | |
| stop | string | array | optional | Up to 4 sequences where generation stops. | |
| frequency_penalty | number | optional | −2 to 2. Discourages repeating tokens. | |
| presence_penalty | number | optional | −2 to 2. Encourages new topics. | |
| seed | integer | optional | Best-effort deterministic sampling. | |
| tools | array | optional | Function/tool definitions the model may call. | |
| tool_choice | string | object | optional | Control tool use: `auto`, `none`, `required`, or a specific tool. | autononerequired |
| response_format | object | optional | Constrain output to `json_object` or `json_schema`. | |
| reasoning_effort | string | optional | For reasoning models: `low`, `medium`, or `high`. | lowmediumhigh |
| stream | boolean | optional | Return SSE deltas instead of one JSON response. | truefalse |
Response fields
A non-streaming call returns a chat.completion object (see the panel on the right).
| Parameter | Type | Required | Description |
|---|---|---|---|
| id | string | optional | Unique identifier for the completion. |
| choices | array | optional | Generated choices. `choices[0].message.content` holds the reply. |
| choices[].finish_reason | string | optional | Why generation stopped: `stop`, `length`, `tool_calls`, or `content_filter`. |
| usage | object | optional | Token accounting: `prompt_tokens`, `completion_tokens`, `total_tokens`. |
| model | string | optional | The model that produced the response. |
Model-dependent features
OneRouter keeps protocol compatibility and forwards model-specific parameters where supported. Features such as tool calling, structured output, reasoning, prompt caching, web search, image input, audio input, video generation, and provider-specific options depend on the selected model and upstream provider. Check the model detail page before relying on a feature in production.