Up to 7.2× the speed of standard llama.cpp Runs 100% on your PC
Free for personal use
Download free
How-to guides/Guide 10
10

Use 5090Ai from other apps (Local API)

4 steps · about 5 minutes

5090Ai can act as a local, OpenAI-compatible server, so other apps, editors and your own scripts can use the same speed.

You'll needA model loaded in 5090Ai
  1. Open the Local API page

    Click API in the menu on the left.

    5090Ai
    The Local API server page, stopped
  2. Switch it on

    Turn the switch to Running. The address to use is shown underneath: http://localhost:1234/v1. Click the copy button to copy it.

    • Serve on local network: lets other devices on your network connect.
    • Require API key: only apps with the key can connect.
    • Allow browser requests (CORS): for web apps that call the Local API from a browser.
  3. Connect your app

    In any app that supports OpenAI, set the base URL to http://localhost:1234/v1. Quick start has ready-to-copy examples for curl, Python and JavaScript. For example, in Python:

    pip install openai
    
    from openai import OpenAI
    client = OpenAI(base_url="http://localhost:1234/v1", api_key="not-needed")
    reply = client.chat.completions.create(
        model="qwen3.6-35b-a3b",
        messages=[{"role": "user", "content": "Write a haiku about GPUs."}])
    print(reply.choices[0].message.content)
    5090Ai
    Local API running, with the endpoints and quick start
  4. See the requests

    The request log shows every call: when, from where, the tokens and the speed.

    5090Ai
    Request log with three requests

    Local API requests count toward the free daily prompts, just like chats.