Google's Antigravity lead Varun Mohan announced the platform has added a new model, Gemini 3.5 Flash (Low), to address user complaints about burning too many tokens on simple tasks. Internal tests show it cuts token generation by about 45% compared to the Medium variant, and on software engineering (SWE) tasks, it actually beats the previous flagship Gemini 3 Flash (High). To make sure developers have plenty of resources to build this week, Antigravity has reset Gemini quotas across all plans.
https://twitter.com/_mohansolo/status/2058740603563966758