Starting 9 October 2026, unpaid Gemini users will lose access to Gemini 3.6 Flash and Gemini 3.1 Pro, leaving Gemini 3.5 Flash-Lite as the only available model.
Subscribers to the $5 monthly Google AI Plus tier will lose access to Gemini 3.1 Pro, requiring an upgrade to the $20 AI Pro or $100 AI Ultra tiers.
Google is implementing low, medium, and high effort controls alongside compute quotas that refresh every five hours up to a weekly ceiling.
Google AI chief architect Koray Kavukcuoglu confirmed Gemini 4 Argon is currently restricted to cybersecurity testers, the US administration, and the Fairwind Program.
Starting 9 October, Google will remove free access to its Gemini 3.6 Flash and Gemini 3.1 Pro artificial intelligence models across web and mobile applications. Free accounts will be restricted entirely to Gemini 3.5 Flash-Lite, the company's least computationally intensive model. Flash-Lite remains functional for basic tasks such as quick internet searches, text translation, summarization, and drafting basic messages. However, the model exhibits lower performance on multi-condition instructions, advanced programming, comprehensive document analysis, and detailed image generation. Paid subscribers on the $5 monthly Google AI Plus tier will also lose access to Gemini 3.1 Pro, retaining only Flash and Flash-Lite.
To maintain access to Gemini 3.1 Pro, users must upgrade to the Google AI Pro plan at $20 monthly (€21.99 in Europe) or the Google AI Ultra plan at $100 monthly (€99.99 in Europe). Google is notifying AI Plus subscribers by email regarding the specific dates their Pro access will conclude. Alongside model reallocations, Google is introducing a three-stage effort selector allowing users to set processing depth to low, medium, or high. Selecting higher effort increases analytical accuracy for complex tasks while consuming available quota faster. Usage allowances are also shifting to computation-based allocations rather than fixed message totals.
“Gemini will transition to compute-based usage limits that will refresh every five hours until reaching the weekly limit.”
Data from the article text
Subscribers to the $100 monthly AI Ultra tier will receive access to an exclusive Deep Think feature and Google's forthcoming flagship model, Gemini 4 Argon. Argon is engineered for advanced reasoning, autonomous programming, financial analysis, and legal research. During internal trials, Argon identified and fixed a critical software flaw in a medical network system that risked leaking confidential patient records. Google reported that other advanced models evaluated during testing failed to detect that specific vulnerability. Google is withholding a wide public release of Argon to evaluate potential misuse scenarios.
Access to Gemini 4 Argon is currently limited to cybersecurity specialists, the United States administration, and participants in Google's Fairwind Program. These trusted partners are tasked with testing system boundaries and identifying operational hazards prior to commercial distribution. Google AI chief architect Koray Kavukcuoglu noted that high-capability systems necessitate a phased rollout strategy. Google has incorporated technical safeguards preventing Argon from assisting with cyberattacks or the development of chemical, biological, and nuclear weapons. The company is actively monitoring system outputs to ensure model behavior aligns with developer safety standards.
Pollar is free and carries no ads in the text. Its members keep it going.
Support PollarAI & Tech · Mountain View
Starting 9 October, Google will restrict free Gemini users to its entry-level Flash-Lite model while removing Pro model access from its entry paid subscription.
AI text based on reports by engadget, wiadomosci.radiozet.pl, La Razón and other publishers. Published .