Nano Banana 2 Lite lands in Leonardo, Magnific, and Figma Weave

• Google shipped Nano Banana 2 Lite into the GoogleDeepMind launch thread as its fastest and cheapest Gemini image model, with OfficialLoganK's launch post putting it at under 4 seconds per image and $0.034 per 1K image.
• Within a day, MayorKingAI's Leonardo post, runwayml's Runway post, and figmaweave's Weave post all announced integrations, which turned a Gemini API model into a same-week design-tool rollout.
• magnific's rollout post framed the model around about 5-second generations, image editing, and up to 14 references per image, while LukeW's consistent-character test said the speed gain came with slightly lower quality than Nano Banana 2.
• Google also positioned Nano Banana 2 Lite as the front end of a two-step media workflow, because GoogleDeepMind's Interactions API demo pairs it with Gemini Omni Flash for image-to-video generation and up to three sequential edits.

You can jump from GoogleDeepMind's linked docs to creator-facing rollouts in Leonardo, Magnific, Runway, and Weave almost immediately. LukeW's hands-on test already surfaced the obvious tradeoff, faster renders that stay close to Nano Banana 2 but do not fully match it, and OfficialLoganK's reply also narrowed the video side of the story to Omni Flash only.

Google's core pitch was speed and cost. GoogleDeepMind's product thread said Nano Banana 2 Lite produces text-to-image outputs in four seconds, while OfficialLoganK's launch post priced it at $0.034 per 1K image and called it Google's fastest and cheapest Gemini image model.

The attached benchmark chart in OfficialLoganK's comparison post gave the broader shape: Nano Banana 2 Lite scored below Nano Banana 2 on image generation and editing Elo, but it cut latency from 20.0 seconds to 4.0 seconds and list price from $0.067+ to $0.034+.

The platform wave hit fast.

• MayorKingAI's Leonardo post said Leonardo added Nano Banana 2 Lite for fast image generation, readable text, precise edits, and mixed image-plus-text guidance.
• runwayml's Runway post said Runway made the model available directly and through Agent.
• figmaweave's Weave thread bundled Nano Banana 2 Lite with Gemini Omni Flash inside Weave, then layered on a 50 percent first-month discount.
• magnific's rollout post pushed the broadest availability claim, calling Nano Banana 2 Lite unlimited on Magnific, MCP, and Spaces.

This is the part creative teams actually care about: the same model showed up in image apps, design tooling, and agent surfaces instead of staying trapped in AI Studio.

Hands-on posts landed on the same theme as Google's chart. LukeW's test said Nano Banana 2 Lite was much faster and very close to Nano Banana 2 for a consistent-character maker, but still not as high quality. halfmage's example showed a text-heavy history fact rendered in about five seconds, which lines up with the model's pitch around quick ideation and readable text.

Magnific made that tradeoff legible in product language. magnific's showcase sold the model as "Pro looks. Lite speed," and magnific's rollout post centered generation time, reference count, and editing flexibility rather than absolute image quality.

The most concrete workflow feature surfaced in partner posts was reference-heavy generation.

• magnific's rollout post said users can attach up to 14 references per image.
• MayorKingAI's Leonardo post emphasized guiding outputs with both image and text references.
• magnific's rollout post and runwayml's Runway post both highlighted image editing alongside plain text-to-image.

That makes Nano Banana 2 Lite look less like a one-shot prompt toy and more like a fast layout, variation, and art-direction engine.

Google's own demos kept pairing the image model with Gemini Omni Flash. GoogleDeepMind's Interactions API demo said the Interactions API can generate an image with Nano Banana 2 Lite, animate it with Omni Flash, and preserve session history across as many as three sequential edits.

Creator tooling picked up the same pattern immediately. magnific's storyboard workflow laid out a three-step flow: generate the first image, create variations in edit mode, then send the storyboard into Seedance 2.0 4K for video.

The video side shipped narrower than the image side. In OfficialLoganK's reply, he said only a Flash version of Omni was available across products, and GoogleDeepMind's launch thread described Gemini Omni Flash, not a broader Omni family rollout.

That matters for how these tools are being packaged right now: Nano Banana 2 Lite is the widely distributed fast image layer, while the paired video story is still the Flash-tier version only.