Gemini 3.2 Flash quietly hit the iOS app. Pricing is the news.

Gemini 3.2 Flash appeared inside the iOS Gemini app and Google AI Studio on May 5 with no press release, no keynote, no launch post. It was first noticed by users seeing model versions cycle in real time on their devices [Build Fast with AI].

── What shipped ──

The model sits between Gemini 3.1 Flash-Lite and 3.1 Pro in the lineup. Reported pricing is $0.25 per million input tokens, with performance close to 3.1 Pro on coding and creative tasks.

Google also pushed Gemini Advanced to a 1M-token context window for paid users, with file upload and document Q&A capabilities arriving in May [Gemini release notes].

── Why it matters ──

The price-per-token is the lede. At $0.25/M input, Gemini 3.2 Flash undercuts every comparable tier from Anthropic and OpenAI. If the quality-versus-pricing claims hold up to independent benchmarking, Google has shipped the cheapest credible mid-tier model on the market.

The quiet rollout pattern is also worth noting. Google has used this approach before — push a model into Arena and the consumer app, see how it benchmarks live, then announce officially when the numbers hold. Expect a formal Gemini 3.2 Flash announcement at I/O 2026.

For developers, the practical action is to A/B-test 3.2 Flash against your current cheap-tier vendor for read-heavy workloads. The 4x price gap versus the comparable Anthropic and OpenAI tiers is large enough to matter at production scale.

── Editor's take ──

Quiet drops mean the team is still measuring. Don't bake 3.2 Flash into a critical path before I/O — pricing or capability could shift on the official launch.