What is this?

A managed API endpoint that forwards your requests to DeepSeek models using standard OpenAI format. If your code works with the OpenAI Python SDK, it works here.

Quick Start

from openai import OpenAI

client = OpenAI(
    base_url="https://modelrelayapis.cc/v1",
    api_key="sk-gateway-YOUR-KEY",
)

response = client.chat.completions.create(
    model="deepseek-v4-flash",
    messages=[{"role": "user", "content": "Hello!"}],
)
print(response.choices[0].message.content)

Same openai package. No new SDK.

Supported Models

• deepseek-v4-flash — Fast and cost-effective
• deepseek-v4-pro — Higher capability for complex tasks

What You Get

• OpenAI-compatible API — Zero code changes beyond base_url and key.
• Streaming — SSE streaming works out of the box.
• Budget-based billing — Pay for API budget. Billed by real token pricing (cache-hit, cache-miss, output).
• Per-key rate limiting — Predictable performance per user.
• Usage dashboard — Check your balance anytime with your key.

Can I use the standard OpenAI Python SDK?

Yes. Set base_url to our endpoint and pass your gateway key.

How is billing calculated?

Billed by actual DeepSeek token pricing: input cache-hit $0.014/M, input cache-miss $0.14/M, output $0.28/M (flash). Pro costs more. Your budget decreases with each request.

What happens when my budget runs out?

You will receive HTTP 402 and your key becomes exhausted. Buy a new package to continue.

Is there a rate limit?

Yes. Each key has a per-minute limit (30 or 60 RPM depending on plan). Exceeding returns HTTP 429.