Up until today I really hated the Antigravity… and 3.8 Flash since they removed 3.6 and 3.7 which were the most balanced models to use saving quota while working faster.
A few factors why I hated Antigravity after they removed 3.6, and 3.7,
- 3.8 was very slow & ate a lot of quota
- Continuously keep reading exact files and line ranges
- Always reading files before making any progress unlike 3.6 and 3.7 which really acted quickly without even reading files most of the times.
Up to today you can check my posts, I have shared almost all the issues and supported discussions related to expressing wrath 
But I really found all the ways to fix those issues and get a use of current Antigravity,
Repeated File-Reading
Actually the real reason it’s keep reading files is related to the mode (Low, Medium, and High). To actually save quota to get a lot of work done, just like we did with 3.6 and 3.7 we do the same with 3.8 flash. But the 3.8 Flash Low has this issue which isn’t really there with Medium and High modes.
The moment I jumped from Low to Medium, it really didn’t keep reading exact same files again and again. So the solution is to skip using Low mode. But this doesn’t justify the thing Google did 
. I think behind all these update to Antigravity has few reasons if we think logically,
-
Antigravity is really a great product which we all would agree even with a lot of gross updates still it works well. So Antigravity’s one of the options a good developer would choose. So Google has already won that title. They no longer need Free users, caz free users have already given what google wants. So they removed all the 3.6 and 3.7 models and are about to remove the MOST IMPORTANT models in it from free plan which is Claude models after November 2, 2026.
-
The other reason I see is that since 3.8 was the latest model before google launched Argon, they started to identify the same issues we did. So you can see that in the Free plan the only model left after November 2 this year is Gemini 3.8 Flash. They can use these Free users who can’t pay and aren’t switch to other platforms because only Antigravity offers this much of free quota fr. So they can use increase the usage of 3.8 Flash with these Free plan users to focus on training it more since it’s the only model left in Free plan.
So from one side Google cuts their costs a lot by removing models from free version and on the otherhand they use the lab rats to train their 3.8 more with intense focus identifying what’s wrong with that model.
Slow & Reading Files always before making any Changes
Actually I see it as a good thing even when it uses a lot of quota. But lately I noticed that the quota is much more higher. Even with a lot of updates and actions, quota is still consumed half. So using the Medium mode and letting it make progress isn’t gonna eat the quota now faster like before.
It’s actually fine now for me. But I have new issues with Antigravity now
,
-
“The server cleared a prefix of the conversation as it grew too large”
This shit is very annoying. I had to roll back to a backup since this shit is shown and couldn’t restore to the checkpoint before everything was broken on my system. At least it should show the history wihle still not using the memory. But actually I tried so many ways and still couldn’t do it…
-
Extremely slow 3.8 Flash
