Nano Banana 2 / Gemini 3.1 Flash Image (Google product announcement)
A product-launch tweet from Logan Kilpatrick (Google AI Studio lead) announcing Nano Banana 2, Google’s next-generation image generation and editing model. Available through AI Studio and the Gemini API under the internal name Gemini 3.1 Flash Image. The announcement also introduces new (lower-cost) output resolutions and a built-in Image Search tool. No technical report, no benchmarks, no weights — this is a closed, API-only release; the page exists as a tracking pointer for Google’s closed frontier image stack rather than as a research artifact.
Key claims
Section titled “Key claims”- Nano Banana 2 is positioned as Google’s “best image generation and editing model” at announcement [tweet body].
- API access is through Google AI Studio and the Gemini API under the model name
Gemini 3.1 Flash Image[tweet body]. - The release introduces new (lower-cost) resolutions and an Image Search tool alongside the model itself [tweet body].
Method
Section titled “Method”Not disclosed. The tweet contains no architectural, training, or evaluation detail — only the product name, the API surface, and a single sample image. Naming convention places it in the Gemini 3.x Flash family, consistent with Google’s strategy of routing image generation through the Gemini multimodal stack rather than through a separate image-only product brand (cf. Imagen).
Results
Section titled “Results”None reported. No benchmark numbers, no comparison table, no example gallery beyond the single attached image. The 579.4K view count is the only quantitative signal on the page and is irrelevant to capability.
Why it’s interesting
Section titled “Why it’s interesting”Nano Banana / Nano Banana 2 / Gemini 3.x Flash Image is the closed-source frontier image-editing baseline that many of the open releases this wiki tracks are explicitly trying to beat — FLUX.2 [klein]: Towards Interactive Visual Intelligence, Qwen-Image-2.0, and HunyuanImage 3.0 Technical Report all benchmark in this neighborhood. Filing the announcement gives future open-release pages a stable pointer for what “Nano Banana 2” actually was at the time of comparison. The Open foundation-model releases concept page already names Nano Banana in its “closed-but-API-accessible” cohort as the comparison baseline open releases position against; this page is the artifact behind that mention.
See also
Section titled “See also”- Open foundation-model releases — Nano Banana 2 is the closed-source baseline against which open image-gen releases position themselves
- FLUX.2 [klein]: Towards Interactive Visual Intelligence — concurrent open frontier image-gen + editing release; targets the same closed competitors
- Qwen-Image-2.0 — Alibaba’s open image-gen update in the same competitive window
- HunyuanImage 3.0 Technical Report — Tencent’s open MoE image-gen release explicitly benchmarking against Nano Banana / Seedream / GPT-Image
- Introducing Recraft V4: Design Taste Meets Image Generation — another closed-API image-gen entrant in the same period
- Gemini 3 Deep Think: Advancing science, research and engineering — same Gemini 3.x family, reasoning side
- Introducing Agentic Vision in Gemini 3 Flash — Gemini 3 Flash multimodal vision capabilities