product update

Google triples Gemini usage limits in Antigravity coding tool twice in one week after user complaints

TL;DR

Google has raised Gemini usage limits in its Antigravity coding tool by 3x twice within one week, responding to developers who hit new compute-based quotas within hours. The company also reset weekly quotas for all paid users twice, though limits remain lower than pre-restriction levels.

2 min read
0

Google triples Gemini usage limits in Antigravity coding tool twice in one week after user complaints

Google has raised usage limits for Gemini models in its Antigravity coding tool by 3x twice this week after developers quickly hit newly implemented compute-based quotas.

The company first tripled rate limits on Wednesday, shortly after new restrictions went live earlier in the week. Users reported hitting their limits within one hour of working in Antigravity, a significant reduction from previous unlimited access.

Google tripled the weekly quota again on Thursday night. Varun Mohan, a Director at DeepMind working on Antigravity, acknowledged users could exhaust their weekly limits "after a couple work sessions." The company has reset quotas for all paid plan users twice this week.

New compute-based limits across Gemini

The usage restrictions were announced alongside multiple Gemini-related updates at Google I/O this week. The new compute-based limits apply across Gemini's various tools and capabilities, not just Antigravity.

Users expressed frustration at the sudden change from effectively unlimited usage to metered access. The quick adjustment in Antigravity suggests Google underestimated how rapidly developers would hit the initial caps during normal workflow.

Limits still lower than before

Despite two 3x increases, usage limits in Antigravity remain below pre-restriction levels, according to user reports. The constraints outside Antigravity remain unchanged at their initial levels.

Google has not disclosed specific numerical limits or pricing details for the compute-based quotas. The company also has not announced whether further adjustments are planned for other Gemini tools experiencing similar capacity issues.

What this means

The rapid double-increase reveals Google misjudged usage patterns for AI-powered coding workflows. Developers working with code completion and generation features consume significantly more model compute than casual chatbot interactions. The episode highlights the challenge AI companies face when transitioning from growth-phase unlimited access to sustainable usage metering—even a 9x increase from the initial limit may not restore previous functionality. Other AI coding tools like GitHub Copilot and Cursor may gain an advantage if Gemini's restrictions remain substantially lower than competitors.

Related Articles

product update

Pixel 11's Gboard Adds Gemini-Powered 'Personalized Suggestions' That Read Screen Context

Google has updated Gboard's Writing tools on the Pixel 11 with a new 'Suggested' tab powered by Gemini Intelligence. The feature drafts personalized message suggestions using typing history and on-screen context, processed via Google's Private AI Compute.

product update

Google Adds AI Study Tools with Interactive 3D Simulations to Search and Gemini

Google announced new study features across Search and Gemini, including AI-generated interactive visuals, 3D simulations, customized practice quizzes, and a dedicated learning hub. The rollout intensifies competition with OpenAI and education startups like Knowt and Gauth for student users.

product update

OpenAI Launches Apple Messages Plugin, Letting ChatGPT Read, Draft, and Send Texts

OpenAI has launched an Apple Messages plugin that lets ChatGPT read, sort, draft, and send texts on a user's behalf. The company says the plugin runs locally and does not index all messages, though it has not detailed exactly how that works.

product update

AWS Bedrock Adds Cross-Region Inference for OpenAI's GPT-5.6 Models

Amazon Bedrock now supports cross-Region inference for three GPT-5.6 variants — Sol, Terra, and Luna — across more than 25 AWS Regions. The feature routes requests to available compute capacity via geographic or global inference profiles, without requiring code changes beyond swapping a model ID.

Comments

Loading...