Google is tightening the screws on its consumer LLM offerings, effectively pushing high-reasoning capabilities behind stricter subscription tiers. Following the compute-based usage changes implemented in May, the company will restrict which Gemini models are available to non-paying and lower-tier subscribers starting October 9, 2026. This move signals a clear segmentation strategy where raw model capability is no longer just a matter of usage limits, but of tiered access.

The New Access Tiers

According to an updated support document, users without a Google AI subscription will be limited exclusively to the Gemini 3.5 Flash-Lite model. They will lose access to both Gemini 3.6 Flash and Gemini 3.1 Pro. For those paying for the AI Plus subscription at $4.99 per month, the restrictions are slightly less severe but still significant: they will retain access to Flash-Lite and Flash, but will no longer be able to use the Pro model. Subscribers in this tier are being notified via email regarding the effective date of these changes.

Pro and Ultra Get More Power

The restrictions on lower tiers are balanced by new features for higher-tier subscribers, with no regressions for AI Pro ($19.99/month) or AI Ultra ($99.99/month) users. In a notable shift, AI Pro subscribers will now gain access to the Deep Think option, previously reserved for the more expensive Ultra plans. This feature, described as offering 'maximum parallel reasoning,' suggests Google is trying to add value to the mid-tier plan even as it strips away the Pro model from the entry-level paid tier. Additionally, the Gemini app is introducing 'low,' 'medium,' and 'high' effort levels for available models, allowing users to trade off between answer thoroughness and usage limits.

The Argon Mystery

The restructuring comes amid uncertainty regarding Google's next-generation frontier model, Gemini 4 Argon. Announced earlier this week, Argon is set to launch exclusively for AI Ultra subscribers first. Google has yet to clarify whether Argon will be classified under the existing 'Pro' tier or if it will spawn a new, higher tier above it. This ambiguity leaves developers and power users guessing about the future architecture of Google's model hierarchy, especially as the company aligns its consumer app features with those found in Google AI Studio and Antigravity.

Key Takeaways

  • Free users lose access to Gemini 3.6 Flash and 3.1 Pro, restricted to Flash-Lite only starting October 9.
  • AI Plus ($4.99/mo) subscribers lose access to Pro models, limited to Flash-Lite and Flash.
  • AI Pro ($19.99/mo) subscribers gain access to Deep Think, previously an Ultra-only feature.
  • New 'effort levels' (low, medium, high) are coming to the Gemini app to manage usage and depth.
  • Gemini 4 Argon's tier classification remains unclear, launching first for Ultra users.

The Bottom Line

Google is aggressively fragmenting its model access to drive subscription upgrades, turning 'Pro' into a premium feature rather than a standard offering for paid users.