Migrating to Tase

If your application already speaks the OpenAI chat completions format, migrating to Tase is a configuration change, not a rewrite. Three values change: the base URL becomes https://api.tase.app/v1, the API key becomes your tase_sk_... key, and the model name becomes tase-8b. Your request-building code, response parsing, and streaming handlers keep working as they are.

Before and after

JavaScript

// Before - an existing OpenAI-compatible setup
import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: process.env.OPENAI_API_KEY,
});

const completion = await client.chat.completions.create({
  model: 'your-current-model',
  messages: [{ role: 'user', content: 'Summarize my day.' }],
});

// After - the same code pointed at Tase
import OpenAI from 'openai';

const client = new OpenAI({
  apiKey: process.env.TASE_API_KEY, // tase_sk_...
  baseURL: 'https://api.tase.app/v1',
});

const completion = await client.chat.completions.create({
  model: 'tase-8b',
  messages: [{ role: 'user', content: 'Summarize my day.' }],
});

Python

# Before - an existing OpenAI-compatible setup
import os
from openai import OpenAI

client = OpenAI(api_key=os.environ["OPENAI_API_KEY"])

completion = client.chat.completions.create(
    model="your-current-model",
    messages=[{"role": "user", "content": "Summarize my day."}],
)

# After - the same code pointed at Tase
import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["TASE_API_KEY"],  # tase_sk_...
    base_url="https://api.tase.app/v1",
)

completion = client.chat.completions.create(
    model="tase-8b",
    messages=[{"role": "user", "content": "Summarize my day."}],
)

What stays the same

• The messages array — system, user, assistant, and tool roles behave identically. Existing prompt and history code needs no changes.
• Tool calling — the tools parameter, tool_calls in responses, and tool_call_id result messages all follow the same wire format.
• Streaming — "stream": true returns the same server-sent event chunks ending in data: [DONE]. SSE parsers and SDK streaming helpers work unchanged.
• Response shape — choices[0].message.content, finish_reason, and the usage token counts are where they always were.

What to check as you migrate

Migration checklist

1. 1

   Create a Tase API key

   Generate a key in your Tase dashboard and store it as TASE_API_KEY in your secrets manager or environment.
2. 2

   Point the client at Tase

   Set baseURL (or base_url) to https://api.tase.app/v1 and swap the key. In most codebases this is one constructor.
3. 3

   Replace the model name

   Search for hard-coded model strings and replace them with tase-8b. If the model name already comes from configuration, change one config value.
4. 4

   Run your existing test suite

   Because the wire format is identical, your current integration tests are the migration tests. Pay attention to any test that asserts on exact model output text.
5. 5

   Verify streaming and tool calls in staging

   Exercise one streamed response and one tool-calling round end to end, and confirm usage numbers are flowing into your cost tracking.