Gemini App Moves Deep Research, Adds Copy Canvas, & More

As we spotted earlier this week, Google is widely rolling out an update to the Gemini model picker that also changes how you access Deep Research.

On gemini.google.com, there are now just four options in the model picker for all users:

2.0 Flash: Fast all-around help
2.5 Flash (preview): Our next reasoning model built for speed
2.5 Pro (preview): Reasoning, math & code
Personalization (preview): Based on your Search history

Google has once again renamed “2.0 Flash with Search history: Personalized to you.” It’s similar to the original name, but this model is now listed as being in “preview” instead of “experimental.”

You now access “Deep Research” from a button in the Gemini prompt bar alongside “Canvas” and “Video” for Veo 2 generation (with Advanced). Tap and enter your research prompt. For paying users, Google no longer distinguishes “Deep Research with 2.5 Pro.”

This started rolling out earlier in the week, and we’re now seeing wider availability across free and paid accounts. Overall, it reflects how Deep Research and Veo 2 aren’t really general models, but rather Gemini capabilities. Meanwhile, those buttons are always visible instead of being hidden behind a dropdown menu.

Google this week highlighted two updates to Canvas. If somebody shares a Canvas with you, “Copy Canvas” (bottom-right corner when you’re signed in) to make your own edits to the code. Additionally, you can “Prompt to make apps that use your camera and microphone.”

Finally, Gemini Advanced subscribers now have access to the GitHub app/Gemini Extension. This ability to “Import code from public or private repositories, and ask questions about it” only shows up on the web.