DeepSeek & Gemini API via Unified Chat Completions
Switch between DeepSeek V4.1, Gemini 4, GLM-5.3, Kimi-k3, and Qwen 3.8 Flash by changing one model string on POST https://deepclub.net/v1/chat/completions.
One OpenAI-Compatible Endpoint for Cost-Effective Text Models
In the live DeepClub catalog, DeepSeek, Gemini text models, GLM, Kimi, and Qwen expose standard POST /v1/chat/completions (GLM-5.3 also exposes /v1/responses). Note that Gemini text models currently use Chat Completions rather than native Gemini REST paths.
import os
from openai import OpenAI
client = OpenAI(
base_url="https://deepclub.net/v1",
api_key=os.environ["DEEPCLUB_API_KEY"],
)
for model_id in ("deepseek-v4.1-flash", "gemini-4-hydrogen", "gemini-4-carbon"):
completion = client.chat.completions.create(
model=model_id,
messages=[{"role": "user", "content": "Translate 'Zero Data Retention' to Spanish."}],
)
print(model_id, completion.choices[0].message.content)
Check live per-million-token rates on Live LLM API Pricing or review Cline & Roo Code Setup.
Frequently Asked Questions
- Should I use a Google Gemini SDK for gemini-4-hydrogen on DeepClub?
- No. Current Gemini text models in the DeepClub catalog expose /v1/chat/completions, so use the standard OpenAI SDK with base_url="https://deepclub.net/v1".
- How do I call DeepSeek from Vercel AI SDK?
- With @ai-sdk/openai 5+, call openai.chat("deepseek-v4.1-flash") instead of openai("deepseek-v4.1-flash") so requests route to /v1/chat/completions.
Updated