I have been critizing Gemini 3.8 Flash since it was released in Antigravity. Not only me, many developers were and are still critizing and frustrated with what Google did to 3.6 and 3.7 only leaving 3.8 in Antigravity. Many of us didn’t the model and antigravity for a few reasons,
- 3.8 Flash was DAMN SLOW.
- 3.8 did more than what we wanted.
- It reads a lot of files before making a change unlike 3.6 and 3.7 which made changes with a few file reads.
- Reading the same file or range again and again.
- Eating up quota faster.
- Many other factors.
But lately after like October 20, instead of Low mode with 3.8 I tried Medium and it really worked. Now the model is really faster than before but not the same speed of 3.6 and 3.7, but really usable now. Also the performance has been improved as I felt while working.
Guys if you are still experiencing the Repeated file reading or slowness try switching from Low to Medium. I didn’t face any of the issues I’ve been critizing before more than anyone else 
.
Even the trust level is higher now. It can really fix issues and bugs without making us frustrate with leaving things to fix again and again. Even the quota is really higher now, that even with Medium mode it doesn’t eat quota faster.
But I must explicitly mention that it’s not in the same speed as 3.6 or 3.7.
Based on my experience, 3.8 Flash had better performance than 3.6 and 3.7. The issues were related slowness, repeated file reading, etc,. right? But I say 3.6 worked for me and would like to see it again still… But Google will not bring them back since they don’t care about users guys. All the top companies and AI companies really don’t care about users more than their profit. So we must find ways to really make things work for us. Isn’t it the truth?
Would like to know if you are really experiencing the issues we faced earlier with 3.8 and Antigravity. Specially for Free and Pro users like ourselves. If you tried switching from Low to Medium, would like to hear if it fixed everything just like it did for me.
Try switching to Medium mode from Low and share your experience with us, so others can do the same if it works for everyone. These were the issues everyone had with 3.8,
- Was very slow.
- Was getting stuck in a file reading loop.
- Was reading the same file or range repeatedly.
- Was eating quota faster.
- Is reading a lot of files before making any changes.
(Still there, but I see it as a good thing since we can fix or implement things with 100% reliability with one shot) - Any other issues to mention you faced in replies.
Check if this solution fixes things for you…
After 1 Day…
Guys, I ran into the repeated file reading issue again and I realized one more thing. While it was happening my internet connection has been connected to a slow WIFI network and that’s when the model became too slow and got into the file reading loop. After sometime I noticed that the model isn’t doing any good, I checked the connection and changed it back to my Fast Internet Connection, and it really worked. So another factor I found for the model slowness and repeated file reading issue is a Slow Internet Connection.
Just wanted to share this as soon as I tested it few times before sharing… Hope you guys could skip the issues by realizing the factors. Share here if you find solutions too…