Short summary
When using an alternative/custom model provider in GitHub Copilot Desktop, requests still consume GitHub Copilot AI quota. Once the Copilot quota reaches 100%, the alternative provider can no longer be used.
Affected version or release
v1.0.10
Installation context
The app is installed locally and configured for a repository workspace. The issue affects AI usage quota accounting when using a custom model provider.
What happened?
I configured an alternative model provider in GitHub Copilot Desktop and expected requests to be handled by that provider, using the provider's own API quota/token limits.
However, GitHub Copilot AI quota continues to decrease while using the custom provider.
Additionally, after the GitHub Copilot quota reaches its limit, the alternative model provider also stops working, even though it has its own available quota.
Steps to reproduce
- Open GitHub Copilot Desktop settings.
- Configure a custom/alternative model provider.
- Select the custom provider as the active model.
- Use AI features such as chat or code completion.
- Check GitHub Copilot account quota usage.
- Observe that Copilot quota is decreasing.
- Exhaust the Copilot quota.
- Try using the alternative provider again.
- Observe that the alternative provider cant be used.
Expected behavior
When a custom model provider is selected, AI requests should use that provider's quota/billing and should remain available regardless of the GitHub Copilot quota status.
Actual behavior
- GitHub Copilot quota is consumed while using a custom model provider.
- When Copilot quota reaches 100%, the alternative provider becomes unusable.
- The custom provider does not appear to operate independently from GitHub Copilot quota.
Additional context
I have attached screenshots showing the issue:
- The custom model provider is configured and selected.
- GitHub Copilot quota usage continues to decrease while using the alternative provider.
- After the Copilot quota reaches its limit, the alternative provider is no longer usable.
Screenshots:
No request IDs or additional logs are available at the moment.
Short summary
When using an alternative/custom model provider in GitHub Copilot Desktop, requests still consume GitHub Copilot AI quota. Once the Copilot quota reaches 100%, the alternative provider can no longer be used.
Affected version or release
v1.0.10
Installation context
The app is installed locally and configured for a repository workspace. The issue affects AI usage quota accounting when using a custom model provider.
What happened?
I configured an alternative model provider in GitHub Copilot Desktop and expected requests to be handled by that provider, using the provider's own API quota/token limits.
However, GitHub Copilot AI quota continues to decrease while using the custom provider.
Additionally, after the GitHub Copilot quota reaches its limit, the alternative model provider also stops working, even though it has its own available quota.
Steps to reproduce
Expected behavior
When a custom model provider is selected, AI requests should use that provider's quota/billing and should remain available regardless of the GitHub Copilot quota status.
Actual behavior
Additional context
I have attached screenshots showing the issue:
Screenshots:
No request IDs or additional logs are available at the moment.