Are local models good enough?

Good Enough is a personal project built to help you find out. It runs local models on a 128 GB AMD Ryzen AI Max+ 395 (Strix Halo) and exposes them through OpenAI- and Anthropic-compatible APIs.

The button generates a free temporary API key that works for one hour. You can use it in the chat here or copy it into your own client. No account or payment is required.

What gets stored?

Good Enough does not persist inference content: your prompts, responses, reasoning, and tool arguments are not saved.

  • The temporary key and chat history exist only in this browser tab; refreshing the page or dismissing the key clears them.
  • If you create an account, the server stores only the records needed for the account, session, and API keys.
  • Anonymous hourly counts of rendered landing views, demo keys, and demo request outcomes are retained as aggregate metrics. They contain no identifiers or inference content.

Optional account

Seven-day API keys and live request timing

A free account lets you create personal API keys that last seven days. It also includes a live view of request timing—TTFT, duration, token counts, processing and generation speed, cache reuse, and capacity. The console does not include inference content and starts empty whenever the page is refreshed. There is no paid tier.

example preview · synthetic events

  1. 12:04:01Z

    request accepted

    chat · admitted

  2. 12:04:02Z

    first output

    TTFT 438 ms

  3. 12:04:03Z

    request complete

    1.8 s · 42 tokens · 38.6 tok/s · cache reused

No prompt, response, reasoning, credential, or username is included.