Overview

The Tase API lets you build applications on top of the same model that powers the Tase assistant. You send messages, the model responds. On top of that base loop you get tool calling for connecting the model to your own functions, and streaming for token-by-token output.

All requests go to a single base URL: https://api.tase.app/v1. Authentication is a bearer token in the Authorization header.

What you can build

• Chat — multi-turn conversations via POST /v1/chat/completions with system, user, and assistant messages.
• Tool calling — describe your functions in the tools array and the model returns structured tool_calls when it wants to use one.
• Streaming — set "stream": true and receive the response incrementally over server-sent events.
• Bilingual output — the model is strong in both Turkish and English, and switches naturally between them.

OpenAI-compatible

The API follows the OpenAI chat completions contract: POST /v1/chat/completions and GET /v1/models. Any SDK or HTTP client that speaks that contract works with Tase — point it at the Tase base URL, pass your Tase API key, and set the model to tase-8b. The tools / tool_calls schema and SSE streaming format are the same too.

Your first request

curl https://api.tase.app/v1/chat/completions \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer $TASE_API_KEY" \
  -d '{
    "model": "tase-8b",
    "messages": [
      { "role": "user", "content": "Say hello in Turkish and English." }
    ]
  }'

Next steps

• Quickstart — get an API key and make your first request in curl, JavaScript, and Python.
• Authentication — key format, headers, and how to keep keys safe.
• Models — capabilities and limits of tase-8b.
• Function calling — connect the model to your own tools.
• API reference — every endpoint, parameter, and response field.