Native DeepSeek Web API. Unlimited tokens. OpenAI compatible.
11 models online → · Recommended: Cline
Live list from API — only models that work right now.
Fast chat — код, боты, IDE, повседневные запросы.
FilesWeb search enabled — актуальная информация из интернета.
SearchFilesFast chat — код, боты, IDE, повседневные запросы.
FilesWeb search enabled — актуальная информация из интернета.
SearchFilesExpert mode — максимальное качество на тяжёлых задачах.
ChatReasoning mode — complex logic, math, multi-step analysis.
ThinkingFilesReasoning mode — complex logic, math, multi-step analysis.
ThinkingSearchFilesReasoning mode — complex logic, math, multi-step analysis.
ThinkingFilesReasoning mode — complex logic, math, multi-step analysis.
ThinkingSearchFilesFast chat — код, боты, IDE, повседневные запросы.
FilesReasoning mode — complex logic, math, multi-step analysis.
ThinkingGET /v1/model-capabilities · 11 supported
Works with your stack
Premium DeepSeek access — clean Web API, no routing surprises.
No token billing. One key — full access for your session.
OpenAI Chat Completions — change base URL and API key.
HTTPS, isolated keys dllm-…
Production proxy. Streaming SSE supported.
POST /v1/chat/completions Authorization: Bearer dllm-••• { "model": "deepseek-chat", "messages": [{ "role": "user", "content": "..." }] }
Chat, reasoning, expert, search — pick the right model.
Same settings everywhere. Cline is our top pick from client feedback.
Cline + deepseek-chat — stable tool calls for coding agents.
import requests
API_KEY = "dllm-XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX"
r = requests.post(
"https://deepseek.llm-api.fun/v1/chat/completions",
headers={"Authorization": f"Bearer {API_KEY}", "Content-Type": "application/json"},
json={"model": "deepseek-chat", "messages": [{"role": "user", "content": "hello"}]},
timeout=120,
)
print(r.status_code, r.json())
For long projects with token balance — use the main LLM API.