DeepSeek-V3.2

引入 DeepSeek Sparse Attention 的正式版, 首个将思考融入工具使用的模型, 同时支持思考模式与非思考模式。

Model ID: deepseek-v3.2 · Type: chat · Provider: DeepSeek

Endpoints: /v1/chat/completions · /v1/messages

Pricing

Capabilities

• reasoning
• function_calling
• prompt_caching
• web_search
• thinking_toggle
• batch_api

from openai import OpenAI

client = OpenAI(api_key="sk-...", base_url="https://api.router.ai/v1")
resp = client.chat.completions.create(
    model="deepseek-v3.2",
    messages=[{"role": "user", "content": "Hello"}],
)
print(resp.choices[0].message.content)

FAQ

How much does DeepSeek-V3.2 cost on 聚合 AI 中转站?

DeepSeek-V3.2 (`deepseek-v3.2`) is billed per usage at $0.3135/1M in · $0.9405/1M out, in USD. Current pricing is always listed at https://yigekey.com/models/deepseek-v3.2.

How do I call DeepSeek-V3.2 through 聚合 AI 中转站?

Send a request to https://api.router.ai/v1/v1/chat/completions with the header `Authorization: Bearer <your API key>` and `"model": "deepseek-v3.2"`. The API is OpenAI-compatible, so any OpenAI SDK works by changing base_url to https://api.router.ai/v1 — no other code change.

Which endpoints does DeepSeek-V3.2 support?

DeepSeek-V3.2 can be called on: /v1/chat/completions; /v1/messages.

What is DeepSeek-V3.2's context window?

DeepSeek-V3.2 accepts up to 98,304 input tokens and can return up to 65,536 output tokens. Requests exceeding the input limit are rejected before reaching the model.

What can DeepSeek-V3.2 do?

DeepSeek-V3.2 supports: function_calling, prompt_caching, reasoning, web_search, thinking_toggle, batch_api.

Who makes DeepSeek-V3.2?

DeepSeek-V3.2 is a chat model from DeepSeek, available through the 聚合 AI 中转站 gateway with the same API key as every other model.

Call it through the 聚合 AI 中转站 OpenAI-compatible endpoint (API base: https://api.router.ai/v1). AI agents can discover and call every model on this gateway through MCP (https://mcp.router.ai/mcp) with no manual integration.

API reference · All models