Google removes model picker for free Gemini app users, locking them to 'Auto' routing that defaults to Flash-Lite

Google has removed manual model switching for free Gemini app users. Users on the free plan "without a plan" now see only one option in the model picker: "Auto" with "Gemini 3 models," according to 9to5Google, which reported the rollout on October 9, 2026.

The change was announced the previous week. 9to5Google says it saw the change on the vast majority of its free test accounts on the evening of October 9 (PT), and that Google appears to be rolling it out quickly.

What changed for free users

• No manual model selection. Free users can no longer choose which Gemini model handles their prompts.
• Auto routing. Google says: "Gemini will use auto model selection to send each prompt to the Gemini model that suits it best. Most prompts will go to Flash-Lite, our fastest model. Prompts that need deeper reasoning may go to Flash or Pro."
• Smart model selection can be turned off. If it is off, chats for users without an AI plan use Flash-Lite for all responses.
• Version labeling is inconsistent. The picker says only "Gemini 3 models." Google's support documentation last week referred to the default as 3.5 Flash-Lite, per 9to5Google.

New thinking levels

Google is also replacing the previous "Extended" option with three thinking levels:

• Low: quick and efficient
• Medium: balanced depth
• High: extra thorough

These match the options in Google Antigravity and AI Studio. They will be available to all subscribers and are still rolling out.

AI Plus tier

Google AI Plus users will "soon" be limited to Flash-Lite and Flash, which the source labels 3.8 Flash. No date for that change has been reported.

What has not been disclosed

• Usage or rate limits for free users under Auto routing
• The routing criteria that decide when a prompt goes to Flash or Pro instead of Flash-Lite
• Context window sizes for the Flash-Lite model in the app
• Whether the app will show users which model answered a given prompt
• Any change to API or AI Studio access. The source does not address it.

This is a change to how the Gemini app packages and routes existing models. No new model has been released.

What this means

This is a cost-control move. Sending most free traffic to the fastest, cheapest model reduces serving costs at consumer scale. Pro-tier compute is reserved for prompts that Google's router judges to need it.

The trade-off is transparency. Free users cannot choose Pro for a prompt they care about, and they cannot easily tell which model produced an answer. Output quality will depend on how well the router classifies difficulty, which Google has not published. Users who relied on picking a stronger model manually will need a paid plan or AI Studio, assuming the latter remains available without restriction.

The Low/Medium/High thinking levels bring the consumer app in line with the reasoning-effort controls developers already see elsewhere in Google's stack. The pending AI Plus restriction to Flash-Lite and Flash suggests Pro access is becoming a differentiator for higher-priced plans. Watch for Google's published limits and routing details, which will determine how large the practical gap between free and paid users becomes.

Source: 9to5Google, October 9, 2026. Quoted statements are Google's, as reported by 9to5Google.