From OpenAI

Migrate from OpenAI to Radium

From OpenAI

Radium implements the standard OpenAI Chat Completions API. Point your existing OpenAI SDK client at Radium and change three values.

Before and after

Python

  from openai import OpenAI
  client = OpenAI(
-     api_key=os.environ["OPENAI_API_KEY"],
-     base_url="https://api.openai.com/v1",
+     api_key=os.environ["RADIUM_API_KEY"],
+     base_url="https://api.radium.cloud/v1",
  )
  response = client.chat.completions.create(
-     model="gpt-5",
+     model="clarke-1.0",
      messages=[{"role": "user", "content": "Hello"}],
  )

TypeScript

  import OpenAI from "openai";
  const client = new OpenAI({
-   apiKey: process.env.OPENAI_API_KEY,
-   baseURL: "https://api.openai.com/v1",
+   apiKey: process.env.RADIUM_API_KEY,
+   baseURL: "https://api.radium.cloud/v1",
  });
  const response = await client.chat.completions.create({
-   model: "gpt-5",
+   model: "clarke-1.0",
    messages: [{ role: "user", content: "Hello" }],
  });

Supported endpoints

Endpoint	Status	Notes
POST /v1/chat/completions	Supported	Full compatibility
POST /v1/responses	Not supported	Coming soon
GET /v1/models	Supported	Lists hal-1.0, clarke-1.0, tycho-1.0
POST /v1/embeddings	Not supported	Use a dedicated embedding provider

Parameter compatibility

Parameter	Status	Notes
model	Required	Use hal-1.0, clarke-1.0 or tycho-1.0
messages	Supported	Same format
max_tokens	Supported	Same behavior
temperature	Supported	Same behavior
top_p	Supported	Same behavior
stream	Supported	Same format
tools	Supported	Same schema
tool_choice	Supported	Same behavior
response_format	Supported	JSON mode and JSON schema
presence_penalty	Supported	Range: -2.0 to 2.0
frequency_penalty	Supported	Range: -2.0 to 2.0
logprobs	Supported	Returns up to 99 logprobs per token
seed	Supported	Integer 0 to 999999 for reproducibility

Known differences

• Model strings are different. Using gpt-* returns a model-not-found error.
• Radium does not publish the names of underlying open-weight models.
• Rate limits and pricing differ. See Rate limits and Pricing.

Model mapping

OpenAI model	Radium equivalent
GPT-5 / o3	Hal 1.0
GPT-5-mini / GPT-4.1	Clarke 1.0
GPT-5-nano	Tycho 1.0