Google just launched a new Gemini 3.5 Flash (Low) model variant for its Antigravity platform to stop simple development tasks from burning through user token quotas. According to an announcement on X by Google’s Varun Mohan, internal testing shows this more efficient configuration generates roughly 45 percent fewer tokens than the Medium version.
Even with the lighter footprint, Mohan noted that the model still beats out the older Gemini 3 Flash (High) on software engineering benchmarks.
The update arrives alongside another full quota reset across both free and paid tiers to make sure developers have enough tokens to build for the coming week. It is clear Google is trying to patch things up quickly.
This fresh announcement comes after a chaotic week for the company’s AI ecosystem. As a matter of fact, just yesterday, we highlighted a statement from CEO Sundar Pichai who promised to improve Gemini’s usage limits following the wave of backlash.
So it looks like the team is scrambling to address the pain points for its userbase, which is always a good sign.
The platform has been in full damage-control mode for days. This follows emergency fixes deployed for the Antigravity 2.0 UI to address user panic after an update accidentally gutted the built-in local code editor and left developers stranded with a bare chat interface.
The trouble originally started building last week during a massive wave of Gemini usage limit backlash right after I/O. Paying Pro and Ultra users found themselves locked out of services for up to five hours after running just a handful of prompts or video generations.
For now, the new Low model tier looks like an aggressive attempt to balance server load with user sanity. Early responses on social media are mixed.
Some developers say the optimization feels like a breath of fresh air, while others argue that lowering the model’s reasoning capability beats the purpose of having a new model to begin with.
Someone in the comments even suggested that Google bring back the tried-and-tested Gemini 3 Flash until Google figures out how to get 3.5 Flash to work better.
While that is unlikely to happen, developers can try building away with the new Gemini 3.5 Flash (Low) model, especially since it comes with a usage limit reset too.



