Google made waves at the I/O 2026 conference with several AI-related announcements, including a significant price cut for its top-tier AI Ultra plan. However, the company quietly implemented stricter usage limits on its AI Pro plan, drawing sharp criticism from users. The changes, communicated via email to subscribers, signal a shift in Google's AI subscription strategy.

The New Credit System

Under the revised Pro plan, Google has abandoned the previous token-based usage limits in favor of a credit system. According to the company, credits are now assigned based on the compute required by each prompt, factoring in complexity, features used, and chat length. While Google claims this approach is fairer, users have reported that a single text prompt can consume up to 13% of their daily quota, sparking outrage on platforms like Reddit, where the update was labeled a 'scam.'

For context, the AI Pro plan originally offered generous token allowances that enabled extended conversations and heavy usage of features like coding assistance, image generation, and file analysis. The new credit system effectively ties consumption to resource intensity, meaning heavy prompts deplete the quota faster. This aligns with a broader industry trend toward more granular usage tracking, but the execution has left many subscribers feeling shortchanged.

Five-Hour Windows and Weekly Caps

Google is also adopting a time-based restriction model popularized by Anthropic. Subscribers now face five-hour windows during which credits can be used, with limits refreshing after each cycle. Once the weekly cap is reached, no further usage is allowed until the next week. This eliminates the flexibility of daily resets and forces users to ration their queries more carefully. The same restrictions apply to other Google AI products like Antigravity and Flow, though the company has not clarified whether context windows are affected.

In addition, Google is discontinuing the 1,000 free AI credits for Flow that were previously offered to Pro subscribers. The company assures users that their experience should not change, but the removal of free credits reduces the overall value proposition. The changes were confirmed via email to Android Authority's managing editor, but Google has not publicly detailed the exact credit limits or token equivalents, stating only that Pro users get four times the capacity of the free tier.

Reactions and Implications

User reaction has been overwhelmingly negative. On Reddit, early adopters reported that a routine coding query consumed an outsized portion of their weekly credits, effectively rendering the plan unusable for power users. Some have accused Google of bait-and-switch, noting that the price for the Pro plan remains unchanged while the service's utility has shrunk. The timing is particularly sensitive because Google only recently reduced the Ultra plan's price to $100 per month—a move many viewed as a bid to push Pro users toward a higher tier.

Industry analysts point out that the move mirrors Anthropic's approach, which also uses compute-based credits and five-hour windows. However, Anthropic has been more transparent about its usage limits, providing detailed metrics in user dashboards. Google, by contrast, offers only a vague 'Usage limit' option in Gemini settings, leaving many to discover the changes only after exceeding their quota. This opacity could erode trust in a market where subscription-based AI services are becoming increasingly competitive.

Historical Context and Market Impact

Google's AI subscription lineup has evolved rapidly since the launch of Gemini Advanced in 2023. The original Pro plan offered unlimited access to Gemini's most powerful models with a premium price tag. Over time, as AI compute costs rose and user bases grew, providers began imposing limits to manage infrastructure strain. Google had been testing weekly limits on certain beta features, and the new credit system represents a formalization of those tests. The company's decision to quietly implement changes without public announcement suggests a deliberate strategy to minimize backlash, but the stealth approach has backfired.

For the broader market, Google's move could set a precedent. If other providers adopt similar compute-based credit models, users may face a future where AI subscription costs are less predictable. The shift also highlights the tension between affordability and performance: as AI models become more capable, the underlying computational demands increase, forcing providers to either raise prices or squeeze existing plans. Google's $50 Ultra price cut may be an attempt to capture budget-conscious power users, but the Pro plan's nerfed limits could drive mid-tier users away entirely.

Even more concerning for Google is the potential reputational damage. The company has long prided itself on user-friendly policies, but the opaque credit system and stealth implementation invite comparisons to the 'enshittification' seen in other tech services. Reddit threads calling for boycotts have already emerged, and some users are migrating to alternatives like Anthropic's Claude or OpenAI's ChatGPT Plus, which have more transparent pricing and usage policies. Google's response—a generic statement that the changes are 'more effective'—has done little to calm nerves.

What Users Can Expect Going Forward

Subscribers who wish to continue using the AI Pro plan should monitor their credit consumption closely. The 'Usage limit' option in Gemini settings now shows real-time remaining credits, though the exact allocation remains undisclosed. Given that Google reserves the right to alter limits without prior notice, users may face further reductions in the future. Those who rely heavily on AI for coding, research, or content creation may find the Ultra plan a more reliable—but costly—alternative.

Meanwhile, Google is likely to refine the credit system based on feedback. The company has not ruled out adjusting the cap for Pro users, but no timeline for improvements has been provided. For now, the message is clear: Google AI's sweet spot is narrowing, and users must adapt to a world where every prompt carries a price. Whether the market will accept this remains to be seen, but the early signs suggest a rocky road ahead.

Source: Android Authority News