I constantly get this issue
Error fetching from https://generativelanguage.googleapis.com/v1beta/models/gemini-1.5-flash:generateContent: [503 Service Unavailable] The model is overloaded. Please try again later.
It’s a server-side issue—check Google’s status page for updates, which mostly remains green
The Ask AI here on forum return similar server side error.
This has been ongoing for over a week. Gemini API is extremely unreliable, you just need to keep retrying until it works. Or choose another provider until reliability improves - it’s good that Gemini is now compatible with OpenAI libraries.
Exactly,
After this, I started using the Gemini 2.0 models, which are stable. At least I did not encounter any overload errors with them. However, they are a bit more expensive than the 1.5 version.
I think the newer models and the lite models are more stable but their responses are not as good as the flash model(by lite models) that’s why I switch to them in outrage. Other providers are too costly!