- Google has completely wiped the quota counters back to zero for all free and paid Gemini users
- The reset accompanies a new iteration of the Gemini 3.5 Flash to fix sudden drop-offs in output quality in Antigravity.
- This update boasts much less and has higher endurance on harder tasks.
If you’ve been tracking Google’s rapid iterations inside Antigravity, you’ll remember that they recently spun up a “Low-effort” variant of Gemini 3.5 Flash. The goal was simple: stop the AI from overthinking basic requests and swallowing your entire token quota on simple coding tasks. However, the execution didn’t quite land perfectly. Google is once again rolling out a new version of Gemini in Antigravity to fix some of the issues that have cropped up since its last iteration.
Varun Mohan, Director at Google DeepMind, working on Antigravity, announced on X that the team has deployed a refreshed version of this Gemini 3.5 Flash model. According to Mohan, this updated variant boasts much less and features significantly higher endurance when tackling harder software engineering tasks.
