DeepSeek, Kimi and GLM are live — one endpoint, one bill.›
Voyai

Migrating from OpenAI

Two lines, in practice.

What changes

  client = OpenAI(
-     api_key=os.environ["OPENAI_API_KEY"],
+     base_url="https://api.voyai.net/v1",
+     api_key=os.environ["VOYAI_API_KEY"],
  )

  resp = client.chat.completions.create(
-     model="gpt-4o",
+     model="deepseek-ai/DeepSeek-V3.2",
      messages=[{"role": "user", "content": "Hello"}],
  )

What stays the same

  • Request and response shapes, including streaming chunks.
  • Tool calling, JSON mode, system messages, multi-turn.
  • Your error handling — same error object shape.

What to watch for

  • Model IDs are different. Use the IDs from the Models page.
  • No organisation header, no project scoping. Isolation is per key.
  • Rate limits are yours to configure per key rather than assigned by tier.