0
Neue Alchemy
Sign inSign upLet’s talk
Signal + NoiseIntelligence, distilled daily.
MODEL SIGNAL · GOOGLE

Nano Banana 2 Lite

A fast, cost-efficient Gemini Image model from Google (Gemini 3.1 Flash-Lite Image / Nano Banana 2 Lite) designed for high-throughput text-to-image generation and editing with latency of roughly four seconds per image.

CATEGORYImage
RELEASEDJune 30, 2026
Key Features
  • Fastest and most cost-efficient Gemini image model in the Nano Banana family, built on Gemini 3.1 Flash-Lite Image (`gemini-3.1-flash-lite-image`).
  • Text-to-image generation latency of approximately 4 seconds for a standard 1K-resolution image, per Google’s announcement.
  • Standard API pricing of about $0.0336–$0.034 per single 1K-resolution image based on 1,120 output image tokens and official Gemini API token rates, not per 1,000 images.
  • Official API/model ID: `gemini-3.1-flash-lite-image`, part of the Gemini 3.1 Flash-Lite architecture.
  • Optimized for high-throughput, low-latency image generation and editing with availability in Google AI Studio, Gemini API, and Gemini Enterprise Agent Platform, plus rollout across consumer products such as AI Mode in Search and the Gemini app.
  • Input pricing of $0.25 per 1M input tokens and output pricing of $1.50 per 1M output tokens for images, as per Gemini API documentation.
  • Supports 1K-resolution image generation (approx. 1024px, ~1MP) with image outputs consuming 1,120 image tokens.

Provider announcement →

Read the Model Signal report →

MODEL SIGNAL

Nano Banana 2 Lite

Google introduces Gemini 3.1 Flash-Lite Image, pushing text-to-image generation into the four-second latency tier.

Bottom line

Released by Google on June 30, 2026, Nano Banana 2 Lite—officially addressable via API as gemini-3.1-flash-lite-image—is optimized for high-throughput text-to-image generation and editing. The model outputs 1K-resolution images in approximately four seconds, establishing a faster, cost-efficient tier within the Gemini image family.

Signal

The primary signal lies in the unit economics and generation speed. According to Google's announcement, the model achieves a text-to-image generation latency of roughly four seconds for a standard 1K-resolution output (approximately 1024px, or ~1MP).

On the pricing front, the Gemini API documentation indicates input costs of $0.25 per 1M tokens and output costs of $1.50 per 1M tokens. Because each 1K-resolution image consumes 1,120 output tokens, the effective standard API cost translates to about $0.0336–$0.034 per single generated image. The model is actively rolling out across multiple Google surfaces, including Google AI Studio, the Gemini API, and the Gemini Enterprise Agent Platform, as well as consumer touchpoints like the Gemini app and AI Mode in Search.

Noise

The dual branding between consumer and developer surfaces may cause initial routing confusion. While the model is publicly discussed under the "Nano Banana 2 Lite" moniker, the official architectural name and API model ID is gemini-3.1-flash-lite-image. Operators must ensure they are pointing to the correct 3.1 Flash-Lite endpoint to realize the stated latency and cost profiles.

Where it fits

Nano Banana 2 Lite is designed for high-volume, low-latency image generation workflows. The operator read is that by cutting generation time to roughly four seconds, this model effectively shifts text-to-image capabilities from asynchronous, background-task queues into synchronous, near-real-time user flows.

The directional implication for developers is a widened aperture for interactive media applications. If the provider's latency and cost facts hold at production scale, this architecture equips teams to embed dynamic image generation directly into chat interfaces, rapid prototyping tools, and high-throughput content pipelines without incurring the heavy UX penalty of standard diffusion wait times.

Model Signal · Signal + Noise · Isaiah Steinfeld