On May 29, 2026 Google announced it had patched multiple usage — accounting bugs in the Gemini app after internal and external reports that some requests were being billed incorrectly. VP Josh Woodward said engineers traced several defects that caused the system to overcount consumption for specific features and deployed corrections to reduce unexpected quota loss. That matters to developers and paying users, including Ultra subscribers who now receive twice as many Omni video generations under the corrected accounting.
One prominent fix addressed Omni video generation, where a bug could allow a single Omni video — or two-to consume a user’s entire video — generation quota. Woodward confirmed the accounting error has been corrected, and Google increased the Omni allotment for Ultra members so they receive double the number of Omni video generations under the new rules.
Google also changed how complex prompts to the Gemini 3.1 Pro model are billed, particularly those that include large files. To prevent runaway consumption, the platform now enforces a capped maximum consumption per prompt for these requests. Woodward emphasized that prompts will still run and produce expected outputs; the new cap limits the reported or billed consumption without altering the result the user receives. Several other billing behaviors were adjusted. Failed requests are no longer charged, removing unexpected deductions when operations do not complete. Flash Lite requests have been set to free, eliminating charges for that class of lightweight calls.
To give developers better insight into where quota is spent, Google expanded consumption reporting at the feature level. Complex features such as Deep Research now present more detailed consumption displays so teams can more easily identify expensive operations and diagnose costly usage patterns. Google also tweaked session and workflow behavior to reduce friction for automation: when a developer selects a specific model, that selection now persists across sessions rather than resetting. Combined with richer consumption displays, this persistence is intended to help teams running automated pipelines or agent — style workflows keep configuration stable and costs more predictable.
These patches arrive days after Google rolled out a revamped Gemini app at I/O with new features, an agent mode, and a revised pricing structure for Gemini AI subscriptions. Google says it will add more transparency around other usage metrics; builders are advised to review quota dashboards and logs, validate expected consumption under the updated rules, and use the improved reporting to tune large — file and complex requests.
Sources
Replies (0)
No replies in this topic yet.
