DeepSeek V4 Flash is the efficiency-focused open-weight MoE text model in the DeepSeek V4 family, with a 1M-token context window, high/max reasoning, structured output, and tool calling. HiAPI exposes this model through the Chat Completions API.
Provider: DeepSeek
Category: text generation
Endpoint: /v1/chat/completions
Status: Available
Cost: --
DeepSeek V4 Flash is the efficiency-focused open-weight MoE text model in the DeepSeek V4 family, with a 1M-token context window, high/max reasoning, structured output, and tool calling. HiAPI exposes this model through the Chat Completions API.
API access
Endpoint
POST /v1/chat/completions
Base URL
https://apidev.hiapi.ai
OpenAI Chat Completions compatible. Use one HiAPI key across all available models.
Input price
280 Credits
/ 1M tokens
Output price
560 Credits
/ 1M tokens
Context
1M
context window
Max output
384K
output tokens
Ask a question to stream the answer and inspect token usage, latency, and estimated cost.
Call DeepSeek V4 Flash with the OpenAI-compatible Chat Completions format. Copy a ready-to-use cURL, Python, or Node.js example below.
/v1/chat/completionscurl -X POST "https://apidev.hiapi.ai/v1/chat/completions" \
-H "Authorization: Bearer YOUR_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-v4-flash",
"stream": false,
"messages": [
{
"role": "user",
"content": "请用三句话解释量子计算，并给出一个生活化的类比。"
}
],
"thinking": {
"type": "enabled"
},
"reasoning_effort": "high"
}'Tip: Replace YOUR_API_KEY with your actual API key from the API Keys page.
Compatible with the OpenAI Chat Completions format. Set the base URL, API key, and model name to get started.
The example enables thinking with reasoning_effort=high. Use max for maximum effort. To disable thinking, set thinking.type to disabled and omit reasoning_effort.
Billing is based on actual input, output, and cached token usage. All prices below are shown per 1 million tokens.
Prompts and context sent to the model
Responses and reasoning generated by the model
These specifications reflect the current public HiAPI text request contract. Undisclosed capabilities are not inferred by the page.
DeepSeek V4 Flash is the efficiency-focused open-weight MoE text model in the DeepSeek V4 family, with a 1M-token context window, high/max reasoning, structured output, and tool calling. HiAPI exposes this model through the Chat Completions API.
HiAPI exposes this model through /v1/chat/completions. Test prompts and parameters in the Playground, then use the same HiAPI API key in your server application.
The Playground and API share the same model ID, so three steps take a tested prompt into production.
Step 1
Test the system prompt, temperature, output length, and reasoning mode in the Playground.
Step 2
View an existing key or create another after signing in; one key works across all available models.
Step 3
Send requests to /v1/chat/completions and track cost with the usage object.
DeepSeek V4 Flash is the speed- and cost-efficient open-weight MoE model in the DeepSeek V4 family. According to DeepSeek, it has 284B total parameters, activates 13B parameters per token, supports a 1M-token context window, and provides non-thinking, Think High, and Think Max modes.
Thinking mode is enabled by default. In the Playground, turn thinking off or choose High for standard thinking and Max for maximum effort. API requests use thinking.type=enabled/disabled and reasoning_effort=high/max. Use max explicitly for maximum effort; xhigh maps to high on DeepSeek V4 Flash.
Send POST requests to /v1/chat/completions, set model to deepseek-v4-flash, and authenticate with a HiAPI API key. Image, video, and audio models continue to use /v1/tasks.
HiAPI bills actual input, output, and cached-input tokens separately. Reasoning generated in thinking mode is model output. The model and pricing pages show the current public rates, while usage and API logs record the final charge.
Keep the Chat Completions messages shape, set base_url to https://api.hiapi.ai/v1, use a HiAPI API key, and set model to deepseek-v4-flash. Add thinking and reasoning_effort when you need thinking mode.