Google triples Antigravity usage limits

Google increased usage limits for Antigravity twice on May 21 after rolling out new compute-based caps across Gemini, according to 9to5Google. The changes effectively tripled the coding tool’s rate limits and reset weekly usage caps for users who had already run into the new restrictions. Google made the move as it was also repositioning Gemini 3.5 Flash as its default model for coding and agentic workloads. ### Why did Antigravity users suddenly hit limits? Google introduced compute-based usage limits for Gemini during its I/O 2026 product wave, and Antigravity users began seeing the effects almost immediately, 9to5Google reported. The publication said users who had not previously treated limits as a major constraint were met with new caps as the company adjusted how usage was measured. (9to5google.com) 9to5Google reported on May 21 that Google then raised Antigravity’s limits twice in quick succession. The article said the result was a threefold increase from the initial post-change level, along with a reset of weekly quotas for affected users. ### What exactly changed inside Google’s coding stack? Google said at I/O 2026 that Gemini 3.5 Flash is “the first in our latest family of models combining frontier intelligence with action,” and described it as central to its newer agent workflows. (9to5google.com) In the same I/O roundup, Google said advances to Antigravity were meant to move beyond tools that help write code toward agents that help users act. (9to5google.com) The Keyword, Google’s official blog, said Gemini 3.5 Flash is its “strongest agentic and coding model yet.” Google said the model outperforms Gemini 3.1 Pro on benchmarks including Terminal-Bench 2.1, GDPval-AA and MCP Atlas. ### Why did Google switch the default to Gemini 3.5 Flash? Google said Gemini 3.5 Flash was built for “agents and coding” and positioned it around speed as well as capability. (blog.google) Yahoo Tech reported that Google said the model runs at four times the speed of comparable frontier models and is often available at less than half the cost. Yahoo Tech reported that Google presented Gemini 3.5 Flash as replacing Gemini 3.1 Pro as the default choice for coding and agentic workloads. (blog.google) In a separate Yahoo Tech analysis, the outlet said the release showed Google shifting its pitch from conversational AI toward systems that can plan, build and iterate across multi-step work. ### How does Antigravity fit into the broader I/O push? (tech.yahoo.com) Google described Antigravity at I/O 2026 as an “agent-first development platform” and said AI Ultra subscribers would receive a usage limit five times higher than the Pro plan in the Gemini app and Google Antigravity. The company also said that plan includes Gemini 3.5 Flash integration for testing, debugging and iteration. (tech.yahoo.com) Google’s May 20 I/O announcements page listed Antigravity, Gemini Omni and Gemini 3.5 among the company’s headline launches. That places the quota changes inside a larger rollout aimed at developer tools and agent products rather than a standalone pricing adjustment. ### What comes next for developers using these tools? Google said Gemini Spark would begin rolling out in beta to AI Ultra subscribers in the United States the week after I/O 2026. (9to5google.com) Yahoo Tech also reported that a larger Gemini 3.5 Pro model was already in internal use and was expected to roll out next month. For now, the immediate next step is operational rather than speculative: developers using Antigravity are working under the higher limits Google restored on May 21, while Google’s current default for coding and agentic tasks is Gemini 3.5 Flash. (blog.google) (9to5google.com 1) (9to5google.com 2)