I’m currently building a site using the Firebase Studio AI tool (the one powered by Gemini), and I’ve been running into a frustrating issue. About every 30–45 minutes, the AI seems to freeze — it stops generating any more responses, and nothing loads until I manually restart the VM. I often need to do this multiple times per session just to keep working.
I’ve already switched over to the Blaze plan hoping that would improve performance or provide a more stable environment, but the issue still persists.
Is anyone else experiencing this? And more importantly — is there anything I can do to make the experience more reliable or dedicated so I’m not constantly getting interrupted? Whether it’s a config setting, billing tier, or some usage best practices — I’m open to any ideas.
According to Joe Clinton another useful prompt is “Compress chat”... “Which lets gemini go through and intelligently pick out what code lines it needs to remember and which parts it needs to summarize. Then your chat becomes 10x smaller, but you don’t lose useful context.”
this has helped me, after deleting the full context, but still retaining the project overview by doing some small prompt engineering. I have no slowdown now.