Using the Responses API | DeepSeek API Docs

Skip to main content

On this page<br>Using the Responses API

To meet the demand for Codex, our API now supports the Responses API format, with the base_url being https://api.deepseek.com.

With a simple configuration, you can use DeepSeek models in Codex.

Integrating DeepSeek Models into Codex​

Please refer to Integrate with Codex.

Calling DeepSeek Models via the Responses API​

# Please install OpenAI SDK first: `pip3 install openai`<br>from openai import OpenAI

client = OpenAI(api_key="", base_url="https://api.deepseek.com")

response = client.responses.create(<br>model="deepseek-v4-flash",<br>instructions="You are a helpful assistant.",<br>input="Hi, how are you?",

print(response.output_text)

Streaming​

Set stream: true to receive the response as a sequence of semantic server-sent events (SSE). Each event carries an event field indicating the event type, and a monotonically increasing sequence_number. The stream ends with a response.completed / response.incomplete / response.failed event — there is no data: [DONE] message.

stream = client.responses.create(<br>model="deepseek-v4-flash",<br>instructions="You are a helpful assistant.",<br>input="Hi, how are you?",<br>stream=True,

for event in stream:<br>if event.type == "response.output_text.delta":<br>print(event.delta, end="")

The full list of events:

EventDescriptionresponse.createdThe first event; the response has been created with status in_progressresponse.in_progressThe response is being generatedresponse.output_item.added / response.output_item.doneAn output item (reasoning / message / function_call / custom_tool_call / web_search_call) starts / completesresponse.content_part.added / response.content_part.doneA content part within an output item starts / completesresponse.reasoning_text.delta / response.reasoning_text.doneIncremental chain-of-thought text / the full chain-of-thought textresponse.output_text.delta / response.output_text.doneIncremental output text / the full output textresponse.function_call_arguments.delta / response.function_call_arguments.doneIncremental function call arguments / the full argumentsresponse.custom_tool_call_input.delta / response.custom_tool_call_input.doneIncremental custom tool call (apply_patch) input / the full inputresponse.web_search_call.in_progress / response.web_search_call.searching / response.web_search_call.completedStatus updates of a server-side web search tool callresponse.completedThe final event when the response completes normally, carrying the full response object including usageresponse.incompleteThe final event when the response is truncated (e.g. reaching max_output_tokens), carrying the full response objectresponse.failedThe final event when the response fails, carrying the full response object with error details<br>Compatibility Details​

This section lists the compatibility details of the DeepSeek API with the Responses API. For the full Responses API format definition, please refer to the official OpenAI API reference.

Top-level Request Parameters​

ParameterSupport StatusmodelSupported. deepseek-v4-flash / deepseek-v4-pro, see Models & PricinginputSupported. String or input item list; at least one of input and instructions is requiredinstructionsSupported. Inserted as the first system messagestreamSupportedtemperatureSupported (range [0.0, 2.0]; no effect in thinking mode)top_pSupported (no effect in thinking mode)max_output_tokensSupportedtop_logprobsSupported (range [0, 20])toolsPartially supported. function / web_search supported; other types ignored, see the Tools table belowtool_choiceSupported. none / auto / required / a specific tool ({"type": "function", "name": ...} or {"type": "web_search"} / {"type": "web_search_2025_08_26"})reasoningPartially supported. effort supported; summary accepted but no summary is generatedtextPartially supported. format fully supported; verbosity accepted but has no effectuserSupported. See Rate Limit & Isolationparallel_tool_callsIgnored (parallel tool calling is always enabled)max_tool_callsIgnoredprevious_response_idNot supported (stateless API)conversationNot supported (stateless API)storeNot supported. The response always carries store: falsebackgroundNot supportedmetadataNot supportedincludeNot supportedpromptNot supportedtruncationNot supported. Requests exceeding the context window return a 400 errorservice_tierNot supportedsafety_identifierNot supportedprompt_cache_key / prompt_cache_retentionNot supported. Context caching is managed automatically, see Context Cachingcontext_managementNot supportedstream_optionsNot supported<br>Unsupported parameters are silently ignored and do not cause errors, so existing Responses API clients can connect without modification.

Input Items​

TypeSupport StatusmessageSupported. Roles user / assistant / system / developer (developer is treated as system); content supports strings and input_text / output_text content parts. Image and file inputs are not supported (input_image parts do not...