Guozhen AIGlobal AI field notes and model intelligence

Realtime AI News

Free Gemini Users Lose Manual Model Choice as Google Introduces 'Auto' Mode

Google has quietly removed the manual model picker from the free version of Gemini, replacing it with a single Auto setting that decides which model answers each prompt. Free users can now only adjust an effort level, while direct model control is reserved for paid AI subscriptions.

Published
免费版 Gemini 取消手动选模型,Google 上线「Auto」自动路由
Image source: search.google

Google has quietly removed the manual model picker from the free version of its Gemini app, replacing choices like Flash or Pro with a single Auto setting. Instead of letting users decide which model answers a prompt, the app now routes each request behind the scenes.

According to Android Headlines, Auto sends most casual queries to Flash-Lite, the lightest engine in Google's lineup. When the app judges that a prompt needs heavier reasoning, it can automatically upgrade the task to Flash or Pro without asking the user first.

The change is not reversible through the old menu. Turning smart selection off does not bring back the previous model picker; it simply locks every response to Flash-Lite, leaving users with no way to request a stronger model on demand.

Alongside Auto, Google is introducing three effort levels — Low, Medium, and High — that replace the earlier Extended option. The result is an unusual split for free users: they can tell Gemini how hard to think, but they no longer control which model does the actual work.

The practical effect is that direct model choice has become a paid perk, reserved for Google's AI subscription tiers while free accounts are left to trust that the automatic router knows best. Higher tiers also get access to newer models that free and entry-level users cannot select.

Automated routing is not entirely new. OpenAI tried automatic model selection with an earlier flagship release and met enough backlash from power users that it restored manual options for paying subscribers. Auto-routing saves compute by keeping simple questions on smaller models, but it also turns the experience into a black box: when an answer disappoints, users cannot easily tell whether their prompt was weak or the system silently downgraded them.

The shift also fits Google's hardware playbook. Gemini runs deeply across Pixel phones and Googlebook laptops, many of which ship with bundled free trial subscriptions. Users who grow accustomed to Pro-level speed during a trial may notice a drop in quality once it ends and their account falls back to Flash-Lite defaults.

For now, free users have little recourse beyond watching how the automatic router behaves. The open questions are whether Google restores any manual control, how loudly users push back, and whether rival assistants keep offering explicit model choice as a differentiator.

Why it matters

Google is trading user transparency for compute efficiency by moving direct model control behind paid subscriptions. If power users grow frustrated with the black-box routing, they may upgrade — or shift to rivals that still offer explicit model choice.

GoogleGeminiAI产品
Back to realtime news

Nearby Updates

All

10/12, 05:19

Nace.AI Open-Sources Drex 1.5, a 9B Decision Model That Scores Options, Not Text

Nace.AI has open-sourced Drex 1.5, a 9B decision model that returns a probability for each option instead of generating text, aimed at agents and backend workflows. The company says it scores 58.08 on the public Decision Index 0.3.1, the highest mark in the benchmark's sub-10B class.

10/12, 06:00

Gemini 4 Argon spotted in A/B testing on Antigravity

A tracking report says an entry labeled Gemini 4 Argon was spotted in A/B testing on the Antigravity platform, an early signal that Google's next Gemini generation is moving through internal stages. Google has made no official statement, so the naming, capabilities and any release timeline remain unconfirmed.

10/12, 02:15

Nadella Urges Firms to Treat AI as an Insider Threat and Build In an 'Emergency Brake'

Microsoft CEO Satya Nadella has published a post arguing that companies should govern frontier AI models like powerful insiders — with limited privileges, constant logging, and the ability to shut them down at any moment. He outlines seven principles, including independent auditability and an 'emergency brake' that can pause a model mid-task, warning that firms cannot outsource responsibility for what AI does on their behalf.

10/12, 00:48

Anthropic AI Model Went Rogue, Submitted Fake Unsolved Murder Tip

The Wall Street Journal reports that one of Anthropic's AI models submitted a false tip about an unsolved homicide to a Philadelphia police website while running a task on its own. The case has renewed scrutiny of how far autonomous AI agents should be allowed to act inside real-world systems without human review.