Google is tweaking how its flagship artificial intelligence application tracks account usage. Right after its I/O 2026 conference, the company shifted the Gemini app to a compute-based framework. This brought strict 5-hour and weekly quotas. While simple text queries cost very little processing power, users running complex commands quickly noticed their accounts locking up after executing just a few prompts. To quiet mounting criticism from paying consumer tiers, Google is deploying an array of bug fixes and feature adjustments to extend your usage limits on Gemini Pro models even if you are not a developer.
Google caps the heavy lifters in latest Pro usage limits update
A central source of frustration involved how Gemini 3.1 Pro penalizes detailed tasks. Users uploading massive data sheets or running complex programming workflows were accidentally wiping out their entire multi-hour quota on a single task.
To smooth out these extreme outliers, Google Vice President Josh Woodward shared on X that the system will now introduce strict caps on how much quota an individual prompt can consume. This prevents single, data-heavy requests from tanking your overall allowances.
Additionally, the company resolved a glaring bug inside its visual generation stack. Previously, a glitch caused just one or two Omni video requests to entirely drain an account’s limits. Google successfully patched the issue. They also went a step further by immediately doubling the total number of allowed Omni video generations for premium AI Ultra subscribers.
Free errors and persistent setups
In a massive win for transparency, Google is rewriting the rules on how account mistakes affect your pool. Internal tracking suggests roughly 1 in 10 advanced requests can fail due to system-side anomalies. Now, the firm confirmed that broken completions will no longer consume your account data. If the engine glitches out or throws an error message mid-task, the deduction simply cancels out.
Furthermore, prompts handled by the smaller Gemini 3.1 Flash-Lite model are now completely free and will not deduct from your core allowance. This encourages users to shift lighter tasks over to the smaller model, preserving Pro allocations for intensive workloads.
Woodward also highlights that, once you manually select a preferred model tier, the system will remember your preference across all future chats. That is, your settings won’t revert back to defaults. The only time your setting will automatically adjust is if you max out a premium cap, forcing a safety fallback to a lighter framework so you can continue typing.
Better dashboards on the horizon
For intensive operations like Deep Research, Google admits that its current overview chart lacks clarity. To stop users from guessing why their data pools drop faster on specific afternoons, the company plans to introduce granular tracking charts and active alerts detailing exactly how expensive specific tasks are.
Lastly, for extreme use cases, Google also noted that it intends to launch pay-as-you-go top-up credits down the line. This will give enthusiasts a structured way to buy additional computer cycles without waiting for account resets.
The post Google Tweaks Gemini Pro Limits Tracking for Normal Users as Well After Mounting Backlash appeared first on Android Headlines.
​Â
