With Pro, 10 minutes of work, and then waiting until next week to continue on the flash model?! Wth?
I really like Antigravity and really hope it’s is just a bug. This all Flash 3.5 situation. But sadly, I don’t have the luxury to wait for them to get their sxxx together.
Looking for an alternative IDE I can use instead, maybe with DeepSeek to make it more price-effective?
Not looking for CLI or Vibe coding agent-only environments, looking for something like the old Antigravity experience (now changed to Antigravity IDE).
btw, really great idea, next-level IQ to change the old Antigravity to a new name and give the old name to the new weird thing.
What can go wrong, right? I guess everyone who thinks left to Anthropic? just kidding, you guys probably went on vacation and left the new first-day junior to push code to production 
Any suggestions from other people in the same situation?
Thank you!
Google’s idea of “smart” here is pretty questionable. They increased token limits, but at the same time reduced us from three separate quotas (one per engine type) to just two, now the new engine and the pro engine share the same quota.
In practice, that’s a major downgrade. Flash 3 barely consumed tokens, so this effectively cuts usable work time by around 80–90%. On top of that, the newer models consume significantly more tokens , easily 10x in real usage, so even the remaining quota gets burned much faster.
So this “x3 tokens” improvement? In reality, it just means instead of coding for 5 minutes, I can maybe stretch it to 15… not exactly a win.
Net result: a severe downgrade. We’re getting next to no value for our money, and hitting limits so quickly that the tool becomes unusable.
It honestly feels like Google is squeezing its paying users here, assuming people won’t notice the trade-offs.
I doubt I’m the only one feeling this way, so I’m curious, where are people migrating their dev workflows? I’m currently testing plain VS Code with DeepSeek as a coding agent.
Cursor is no longer an option for me since they now require payment just to use your own LLM API.
If you’ve moved to another IDE or setup, I’d really appreciate hearing what you’re using and the pros and cons.
My Opus Limits are being hit in 1 prompt and having to wait 6 days for it to reset this is awful lol, which IDE did you switch to ?
Right now I am testing vs code using deepseek API , but I am always open to other options as long as they are fair and not Tring to blackmail me to pay the ide creator more money to use it like Google just did (it’s obvious they try to get us to pay 5x if not more to keep using it as we did so far.) have you found any other alternative?
Right now i am researching of Open source IDE + A small Coding Model like 14B or smth and i will be trying to improve it with skills, i saw Qwen 3.6 benchmark tests i guess with the right tools and access to internet i can figure out smth will keep you updated on what i achieve 