Quickstart

Send your first request to a Prism model with an OpenAI-compatible client.

Prism accepts OpenAI Chat Completions requests at https://api.prisminference.com/v1. You need a Prism API key and a public model ID.

Ask the workspace owner for an existing API key, or talk to sales for access.

Send your first request

1. Set your API key

   export PRISM_API_KEY="YOUR_PRISM_API_KEY"
   

   Store the key in your server-side secret manager. Do not commit it or expose it to browser code.

2. Call a model

   cURL

   curl --fail-with-body "https://api.prisminference.com/v1/chat/completions" \
     -H "Authorization: Bearer $PRISM_API_KEY" \
     -H "Content-Type: application/json" \
     -d '{
       "model": "deepseek-v4.1-flash",
       "messages": [
         {
           "role": "user",
           "content": "Write a TypeScript function that limits concurrent promises."
         }
       ]
     }'
   

   Install the OpenAI client with npm install openai or pip install openai.

3. Read the completion

   The endpoint returns a standard Chat Completions response:

   {
     "id": "chatcmpl_01J...",
     "object": "chat.completion",
     "created": 1788998400,
     "model": "deepseek-v4.1-flash",
     "choices": [
       {
         "index": 0,
         "message": {
           "role": "assistant",
           "content": "Here is a concurrency limiter..."
         },
         "finish_reason": "stop"
       }
     ],
     "usage": {
       "prompt_tokens": 18,
       "completion_tokens": 142,
       "total_tokens": 160
     }
   }
   

   Assistant text is in choices[0].message.content. Token counts are in usage.

Stream tokens

Pass stream: true to receive incremental Chat Completions chunks:

const stream = await client.chat.completions.create({
  model: "deepseek-v4.1-flash",
  messages: [{ role: "user", content: "Explain optimistic locking." }],
  stream: true,
});

for await (const chunk of stream) {
  process.stdout.write(chunk.choices[0]?.delta?.content ?? "");
}

Use the Responses format

The same OpenAI client can call the stateless Responses endpoint:

const response = await client.responses.create({
  model: "deepseek-v4.1-flash",
  input: "Explain optimistic locking.",
  store: false,
});

console.log(response.output_text);

See the Responses API reference for input items, function calls, streaming events, and unsupported stateful features.

Use the Anthropic format

Prism also accepts Anthropic Messages requests. Point an Anthropic client at https://api.prisminference.com without the /v1 suffix:

import Anthropic from "@anthropic-ai/sdk";

const client = new Anthropic({
  apiKey: process.env.PRISM_API_KEY,
  baseURL: "https://api.prisminference.com",
});

const message = await client.messages.create({
  model: "deepseek-v4.1-flash",
  max_tokens: 1024,
  messages: [
    { role: "user", content: "Write a tiny rate limiter in TypeScript." },
  ],
});

Next steps

Choose a model

Compare the public model IDs and context windows.

Stream responses

Handle chunks, usage, disconnects, and retries.

Call tools

Run the OpenAI function-tool loop.

Return JSON

Constrain model output to valid JSON or a schema.