Skip to content
HiAPI Docs
HiAPI
Playground Get API Key
English
English 中文

DeepSeek V4.1 Flash Preview API

POST Base URL: https://api.hiapi.ai /v1/responses

The OpenAI-compatible Responses endpoint is recommended for this model, while /v1/chat/completions remains available for compatibility. The same HiAPI API key works across enabled models; media models use /v1/tasks and a different request shape.

Model overview

Model namedeepseek-v4.1-flash
VersionPreview
TypeText generation · Responses + Chat Completions
Reasoningnone / high / max
StreamingSupported
PricingHiAPI live pricing

DeepSeek V4.1 Flash Preview is available through HiAPI for text generation, reasoning, and streaming. Use Responses for new integrations or retain Chat Completions for existing clients.

Production guidance

Stateless requests
  • Send required conversation history in input on every request.
  • Omit store. Do not send previous_response_id, conversation, background, or context_management.
  • Use the same HiAPI API key across enabled models; media generation uses /v1/tasks and a different request shape.

Best suited for

Text responses

Generate text with an explicit reasoning effort and output budget.

inputreasoning.effortmax_output_tokens

Request parameters

model string required

Use deepseek-v4.1-flash.

example deepseek-v4.1-flash
input string | array required

Text or message items; replay the required conversation each turn.

stream boolean optional

true returns semantic SSE events.

default false
max_output_tokens integer optional

Output budget including reasoning tokens. The example is not the model limit.

example 128
reasoning object optional

Examples explicitly select high; choose an effort for your request.

effort enum optional

none disables reasoning; high and max enable reasoning.

enum: nonehighmax

API examples

Request examples

Disable reasoning

Choose none for a response without reasoning.

Request body
{
  "model": "deepseek-v4.1-flash",
  "input": "Reply with OK only.",
  "reasoning": {
    "effort": "none"
  },
  "max_output_tokens": 128,
  "stream": false
}
Max reasoning

Select the verified max effort explicitly; increase the output budget for longer work.

Request body
{
  "model": "deepseek-v4.1-flash",
  "input": "Reply with OK only.",
  "reasoning": {
    "effort": "max"
  },
  "max_output_tokens": 128,
  "stream": false
}
Streaming

Read semantic SSE events and the terminal response usage.

Request body
{
  "model": "deepseek-v4.1-flash",
  "input": "Reply with OK only.",
  "reasoning": {
    "effort": "high"
  },
  "max_output_tokens": 128,
  "stream": true
}

Response schema

Read output_text items from output for non-streaming calls and process semantic event types when streaming. usage is the billing detail source.

{
  "id": "resp_example",
  "object": "response",
  "status": "completed",
  "model": "deepseek-v4.1-flash",
  "output": [
    {
      "id": "msg_example",
      "type": "message",
      "role": "assistant",
      "status": "completed",
      "content": [
        {
          "type": "output_text",
          "text": "Sparse attention computes only selected token relationships."
        }
      ]
    }
  ],
  "usage": {
    "input_tokens": 24,
    "output_tokens": 18,
    "total_tokens": 42,
    "input_tokens_details": {
      "cached_tokens": 0
    }
  }
}
  1. Read output_text content from output for a non-streaming response.
  2. Read usage.input_tokens, usage.output_tokens, and usage.total_tokens; output_tokens_details.reasoning_tokens identifies reasoning usage when present.
  3. In a stream, read the response usage from the terminal event. Handle incomplete or failed responses explicitly.

FAQ

What does Preview mean?

This model is offered as DeepSeek V4.1 Flash Preview. Use the public model ID shown here and verify behavior against your workload when the model version changes.

Which endpoint and model ID do I use?

Use model deepseek-v4.1-flash with POST /v1/responses. POST /v1/chat/completions remains available, with messages instead of input. An OpenAI SDK client can use the HiAPI base URL https://api.hiapi.ai/v1 and its HiAPI API key.

How do I select reasoning?

Responses uses reasoning.effort: none, high, or max. The examples explicitly use high; none disables reasoning. Do not assume effort aliases from other models apply.

How do I continue a conversation?

Responses is stateless for this model. Replay required user and assistant messages in input on every turn. Omit store and do not send previous_response_id, conversation, background, or context_management.

How does streaming finish?

Append response.output_text.delta events and read usage from response.completed. Handle response.incomplete and response.failed as terminal outcomes too. Do not wait for the Chat Completions [DONE] marker.

How are tokens billed?

Input, output, and cached-input tokens have separate rates. Reasoning tokens consume the output budget; inspect usage and the live HiAPI price rather than hardcoding a price from this guide. View live pricing.

Are images, tools, and structured JSON supported?

These capabilities have not been verified for this HiAPI integration and are not advertised on this page. Use text input for the documented request contract.

Next steps