From Rolling Cooldowns to Weekly Gemini Usage Limits
Google is shifting Gemini from flexible, replenishing limits to stricter plan-based caps, and users are feeling the squeeze. Previously, most people saw Gemini as a meter that refilled every few hours or daily, letting them resume heavy use after a brief cooldown. Now screenshots shared from the Gemini app show language about “plan limits” that determine how much you can use Gemini over time, with quota bars and weekly-style tracking instead of purely short rolling windows. For free AI tools, that is a major behavioral change: burn through your allowance in a weekend and you might be effectively locked out for days. Google’s own support pages now warn that Gemini usage limits may change frequently and can be tightened during testing or peak demand, as the company experiments with more aggressive ways to throttle heavy usage while nudging power users toward paid AI Pro subscriptions.
Paid AI Pro Subscribers Run Into New Token Limits
The frustration is not limited to free users. Following Google I/O 2026, paying AI Pro and AI Ultra subscribers began reporting that new compute-based token limits were drastically shrinking how much work they could do in a single session. Where Pro and Ultra once enjoyed roughly 33x and 166x the free Gemini quota, users now say those multipliers have collapsed to around 4x and 20x, with some hitting the ceiling after just a handful of prompts or a few video generations. When the quota is exhausted, subscribers face multi‑hour lockouts before limits refresh. Under the revised system, Google grades each request by complexity and context length, meaning long research threads or intensive coding sessions can burn through allowances shockingly fast. For power users who adopted AI Pro to replace traditional tools, these token limits make a supposedly premium service feel barely more capable than the free tier.
Gemini Now Feels As Restrictive As Claude for Heavy Workflows
Google’s move mirrors a broader industry trend where AI access is metered by compute rather than simple message counts, similar to how Claude enforces its own caps. With Gemini, the new usage model combines five‑hour refresh windows and overarching weekly quotas that throttle intensive workflows. Tasks like using Gemini 3.1 Pro for large‑context research, complex coding, or chains of multi‑modal prompts can chew through a week’s allocation in under an hour, according to early reports, triggering long cooldowns. This has sparked a wave of complaints that Gemini usage limits now match or even undercut competing systems in strictness, while some testers say the newer Gemini 3.5 Flash model feels less reliable than the older 3.1 Pro they’re being pushed away from. The result is a growing perception that both free AI tools and paid plans are converging on tightly controlled, time‑boxed access rather than the always‑on assistants many users expected.
Antigravity Quotas Raised After Backlash, But Wider Caps Stay
The most intense backlash has come from developers using Antigravity, Google’s AI coding environment built on Gemini. After the company quietly nerfed AI Pro limits, Reddit threads filled with reports from paying customers hitting weekly ceilings after just a couple of deep work sessions. In response, Google DeepMind director Varun Mohan announced two rapid quota boosts for Antigravity: first tripling Gemini rate limits and resetting weekly quotas, then tripling them again. In practical terms, Antigravity users now enjoy roughly 9x the post‑nerf limits, and Google says it saw a surge of building activity as soon as the first increase landed. Crucially, though, these improvements apply only inside Antigravity; broader Gemini usage caps across the web and app remain unchanged. For many subscribers, that partial rollback underscores a sense that Google is still calibrating how far it can push monetization without driving users away.
What Google’s Limits Reveal About the Future of ‘Free’ AI
Behind the controversy is a simple reality: running large AI models is expensive, and the era of effectively unlimited free AI access is closing. Google’s weekly Gemini usage limits for free users and stricter token limits for AI Pro subscribers show a clear shift toward treating AI like metered infrastructure rather than a flat‑rate software feature. The company is layering on perks such as YouTube Premium Lite and new creative tools to make paid tiers more attractive, while experimenting with dynamic throttling that can tighten during periods of high demand. For users, the new regime forces a reassessment of what “free” and “pro” really buy: occasional, lightweight assistance versus guaranteed, sustained access for serious work. As competitors adopt similar strategies, the debate is no longer just about model quality, but about whose quotas, refresh windows, and pricing structures best match the workflows people are building around AI.
