Gemini adopts Claude-like limits, frustrating users with disrupted workflows

At a glance:

• Google's Gemini now uses a compute-based limit with a five-hour window, mirroring Anthropic's Claude.
• Users report hitting limits faster due to hallucinations and errors in Gemini's outputs.
• Claude offers superior third-party integrations and interactive visuals despite lacking multimodal capabilities.

The shift to compute-based limits

Google's Gemini has introduced a significant change to its usage model, moving from a prompt-based limit to a compute-based system that includes a five-hour window for both free and paid users. This mirrors a feature previously criticized in Anthropic's Claude, which imposes strict time-based restrictions during peak hours. While Google claims the new limits help manage demand, users argue that the company's vast resources make such constraints less justified compared to smaller competitors like Anthropic.

The five-hour limit has already disrupted workflows for many users. Unlike Claude, where hitting the limit simply pauses access until the next window, Gemini's implementation counts failed or incorrect outputs toward the total. For example, generating multiple incorrect Pokémon Go event visuals consumed nearly a third of one user's limit without producing a usable result. Similarly, using Gemini Canvas to create a mini-app required additional prompts to correct an initial spreadsheet output, further depleting the allowance.

Quality and reliability concerns

Despite its multimodal capabilities and integration with Google's ecosystem, Gemini struggles with accuracy in critical tasks. Users report frequent hallucinations, where the chatbot confidently generates incorrect information. In one test, Gemini misstated dates and invented events when creating Pokémon Go schedules, while Claude made only a single minor error. When processing spreadsheet data, Gemini underestimated totals and defended its flawed calculations, whereas Claude provided accurate results.

These reliability issues are compounded by Gemini's inability to self-correct effectively. Pointing out errors often leads to argumentative responses or repeated mistakes, forcing users to abandon tasks. The chatbot's confidence in incorrect outputs wastes compute resources, making the new limits feel punitive for users seeking accurate results.

Integration and ecosystem limitations

While Gemini integrates deeply with Google services like Gmail, Docs, and Keep, it lacks meaningful third-party support. Users can connect to Canva and Verify AI, but these options are limited compared to Claude's broader ecosystem. Claude supports nuanced permissions for integrations, allowing read-only access to tools like Asana and better synchronization with shared Google Calendars. Gemini, however, remains confined to a walled garden, restricting its utility for users who rely on non-Google productivity apps.

Claude's interactive visuals also provide functionality absent in Gemini. For instance, recipe generation creates an interactive card with timers and cooking mode, while sports scores and weather data are presented in user-friendly formats. Custom charts and timelines further enhance its versatility, offering features that Gemini's current implementation lacks.

Strategic implications for Google

The shift to stricter limits reflects Google's aggressive push to monetize Gemini through paid subscriptions. By embedding AI features into core services like Search and Photos, the company risks alienating users who value free, unrestricted access. The inability to disable AI Overviews (which don't count toward limits) while AI Mode does highlights a potential bait-and-switch strategy, encouraging adoption before introducing paywalls.

This approach contrasts with Claude's more transparent model, where free users face clear time-based restrictions. Google's strategy may backfire if users perceive the limits as arbitrary or if competitors continue offering better value. The article underscores growing skepticism about Google's AI expansion, particularly among users concerned about data privacy and long-term feature availability.

What to watch next

The effectiveness of Gemini's new limits will depend on future adjustments. Google has acknowledged initial bugs causing rapid limit exhaustion, but the five-hour window remains. Users will likely continue favoring Claude for high-stakes tasks until Gemini addresses reliability and integration gaps. Meanwhile, the broader AI chatbot market may see increased competition as companies balance resource constraints with user expectations.

Conclusion

While Gemini's multimodal capabilities and Google ecosystem integration offer potential advantages, its recent adoption of Claude-like limits and persistent accuracy issues have frustrated users. The compute-based model penalizes errors that consume resources without delivering value, while limited third-party support restricts its utility. As Google pushes deeper into AI monetization, the chatbot's success may hinge on addressing these shortcomings and rebuilding trust with its user base.