Google Accidentally Unveils Gemini 3.2 Flash Ahead of Google I/O

Screenshots circulating on the Internet have revealed the existence of Gemini 3.2 Flash within the Google app, even before an official announcement. This discovery comes just ahead of Google I/O, scheduled for May 19 and 20, where Google is expected to showcase its latest innovations.

The screenshots display an option titled "Full Assistance," positioned alongside the existing models Gemini 3.1 Lite and Pro. Although Google has not yet confirmed the information, this leak is already generating significant interest.

Performance and Pricing of Gemini 3.2 Flash

Initial feedback from testers indicates that Gemini 3.2 Flash could compete with heavier versions, particularly for creative and technical tasks. Some users have mentioned generating urban landscapes in ASCII and highly accurate SVG files. One tester even claimed that the model would be "almost equivalent" to Gemini 3.1 Pro for certain tasks.

In terms of pricing, leaks from Google AI Studio suggest a cost of $0.25 per million tokens for input and $2 for output. The knowledge base for this model would be cut off in January 2026, with a potential launch on May 5.

Strategic Stakes for Google

If the release of Gemini 3.2 Flash is confirmed, it fits into Google's strategy of rapidly developing the Gemini lineup. The goal is to offer increasingly powerful models and enhance the so-called "agentic" capabilities of AI, meaning its ability to act autonomously and contextually.

Expectations are high for this new version, which could provide smarter assistants and more natural interactions. For now, Google has not commented on these leaks, but Gemini 3.2 Flash is already shaping up to be one of the most anticipated surprises of Google I/O.