DeepSeek API
Reasoning-first LLMs via RunAPI — flash for fast, low-cost work; pro for complex agentic tasks.
# Base URL
https://runapi.ai
# Endpoints
POST /v1/chat/completions
POST /v1/responses
POST /v1/messages
POST /v1beta/models/{model}:generateContent
POST /v1beta/models/{model}:streamGenerateContent
curl https://runapi.ai/v1/chat/completions \
-H "Authorization: Bearer $RUNAPI_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "deepseek-v4-flash",
"max_tokens": 1024,
"messages": [
{
"role": "user",
"content": "Refactor this Python module for readability, explain each change, then add unit tests for the edge cases."
}
]
}'from openai import OpenAI
client = OpenAI(
base_url="https://runapi.ai/v1",
api_key="your-runapi-key"
)
response = client.chat.completions.create(
model="deepseek-v4-flash",
messages=[{"role": "user", "content": "Refactor this Python module for readability, explain each change, then add unit tests for the edge cases."}]
)import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://runapi.ai/v1",
apiKey: "your-runapi-key"
});
const response = await client.chat.completions.create({
model: "deepseek-v4-flash",
messages: [{ role: "user", content: "Refactor this Python module for readability, explain each change, then add unit tests for the edge cases." }]
});DeepSeek is a family of reasoning-first language models. deepseek-v4-flash is the fast, low-cost tier with an optional thinking mode; deepseek-v4-pro targets the hardest reasoning and agentic workloads. Both are available through RunAPI with one key and per-token billing.
- 多個變體，對應不同速度／品質層級
- Model skill 包含文件、schema 與 setup 備註
- 支援 app-focused coding workflows
- 生成失敗不收費
比較所有 API 變體
DeepSeek API 端點
使用 OpenAI 或 Anthropic SDK 搭配 RunAPI key 呼叫，不需額外 SDK。
| Endpoint | Protocol |
|---|---|
| /v1/chat/completions | OpenAI compatible |
| /v1/responses | OpenAI Responses |
| /v1/messages | Anthropic compatible |
| /v1beta/models/{model}:generateContent | Gemini generateContent |
| /v1beta/models/{model}:streamGenerateContent | Gemini streamGenerateContent |
從 model skill 到第一次結果，只需四步
選擇模型
挑選符合輸出類型、品質門檻與延遲目標的模型與變體。
設定
設定 RunAPI key，並在 coding workspace 安裝 model skill。
開發
使用 skill 指引，在你的 app 內加入模型功能。
接收
透過 task ID 查詢、在支援時串流，或處理 webhook callback。
DeepSeek API 是甚麼？
DeepSeek models are large language models tuned for reasoning, code, and agentic tool use. Through RunAPI they share a single API key with pay-as-you-go token billing, and are callable from both the OpenAI and Anthropic SDKs.
為何透過 RunAPI 使用 DeepSeek API
一個驗證，對應所有 provider
一組 RunAPI key 即可解鎖整個 catalog。無需分開帳戶，亦無需為每個整合輪換密鑰。
統一定價與計費
按次以 USD 收費，每月結算。失敗的生成不會收費。
內含 schema 的 skill
型別化 schema 與 setup 備註打包在 model skill 內，讓實作從正確契約開始。
常見問題
What is the difference between deepseek-v4-flash and deepseek-v4-pro?
Flash is the fast, low-cost tier for everyday tasks; pro targets the most complex reasoning and long agentic workflows at a higher token price.
How do I turn on thinking (reasoning) mode?
Thinking is a request parameter on deepseek-v4-flash, not a separate model — set it per request when you want step-by-step reasoning.
Can I call DeepSeek with both the OpenAI and Anthropic SDKs?
Yes. DeepSeek is available on both the OpenAI Chat Completions and the Anthropic Messages surfaces; point either SDK at RunAPI and pass the deepseek model id.
How are cached prompt tokens billed?
Cached prompt tokens bill at a much lower cache-read rate than fresh input tokens; cache hits are detected automatically and reflected in usage.
Which DeepSeek variants should I use?
Use deepseek-v4-flash for fast everyday work and deepseek-v4-pro for more complex reasoning and agentic workflows.
我應該先用邊個版本？
先選最平而且符合你質素要求的版本。大多數團隊會先用快速版本，之後再升級到 pro 用於正式生產。
有免費方案嗎？
新帳戶可獲得每個模型的首次呼叫免費。之後按次收費。
你哋支援串流結果嗎？
如有提供串流，RunAPI 會端到端串流。
失敗會點樣收費？
生成失敗不會收費。
輸出有快取嗎？
已生成輸出會儲存，並可按任務 ID 取回。輸入內容不會快取。
可以商業使用嗎？
可以——除非模型授權明確限制，否則每個版本都包括商業使用。相關限制會在版本頁面列明。
限流點樣計？
每個密鑰的限流會跟使用等級而提升。最新限制請參考定價頁。
可以在邊度報告問題？
可以在公開 GitHub repo 開 issue，或者電郵支援。