A useful next step

Get your setup working.

Check the app version, identify the model route and capture one reproducible problem. These steps help you get to a useful answer faster.

Start with the right guide

Common problems

What you seeWhat to checkNext step
The app has no model replyIdentify whether you selected local, self-hosted or API inference. Confirm the configured model exists and the runtime or server is running.Try one short prompt. Record the exact error and selected route before changing settings.
Local or server connection failsCheck the endpoint in Settings against your runtime's documented address. Confirm any required tunnel is active and that the runtime responds.Use the project's setup guide. A website account does not start a local model or replace your server.
API authentication or model errorCheck the provider, endpoint and exact model ID. Confirm your provider account allows that model and your key is valid.Re-enter the key in the app's credential settings. Keep it out of screenshots, public issues and logs.
Rate limit or quota reachedRead the provider's current usage and billing limits. A free allowance may vary by account and model.Wait for the provider's reset, reduce requests or explicitly select a configured alternative. An alternative provider receives the task context sent to it.
A long task seems idleLook at the last activity, elapsed time and any visible run controls. Local inference can take longer on slower hardware.Use Stop or Steer where available. Try a smaller task and inspect the resulting files before continuing.
Store and website versions differCheck your installed version and installation channel. A prepared update may still be awaiting publication or certification.Use Microsoft Store for a Store edition, or the documented release route for a direct edition.
A website link is brokenCopy the complete URL and note whether the failure is DNS, a missing page or an unavailable external service.Email the URL and expected destination. Start again from the download directory or guides.

Send a useful issue report

For reproducible software bugs, use the project's issue tracker. For private or security-sensitive questions, email shaf@talktoai.org. Include only the information needed to explain the problem.

Copy this checklist

  • App name, version and Store or direct edition
  • Windows version and relevant hardware
  • Selected model route and model name, without credentials or private server details
  • Exact steps to reproduce with a small, non-sensitive example
  • Expected result and actual result
  • Exact error text, with personal data and secrets removed
  • Whether restarting the runtime or trying a short prompt changes the result

Your data and model choice

A local model route and an external API route have different data flows. Before enabling a provider, check what the app sends, the provider's terms and the relevant product privacy notice. This website is an information hub; it does not provide an upload form for private documents or run your desktop model.

TalkToAi Code privacy · ZERO ONE privacy · ZSEC privacy · ZERO Ouroboros privacy · Website privacy