Adding Nano Banana 2 Lite Support
Generated by Nano Banana 2 Lite — the actual test run for this update.
Quick update on the nano-banana CLI: Google shipped Nano Banana 2 Lite on June 30 — its fastest, cheapest image model yet — and it's now wired in as the new default.
What changed
- Added
--model lite, mapped togemini-3.1-flash-lite-image - Made it the default model (was model 2)
- Output resolution is fixed at 1K (the only resolution Lite supports); the tool now warns if you request anything else
- Updated the cost calculator,
infocommand, and README with confirmed pricing
The numbers
Verified against Google's own DeepMind and Cloud docs pages, since a wave of near-identical explainer posts had already appeared within days of the release:
- $0.0336 per image at 1K
- ~4 seconds per generation
- ~2.7x faster than Nano Banana 2
Ran one real generation to confirm end-to-end — the hero image above:
nano-banana generate "a raccoon holding a sign that says hello lite"Reported cost: $0.0336 — matched the published pricing exactly.
Related posts
This tool keeps pace with Google's release cadence:
- Adding nano-banana-3 Support (Nov 2025) — added Model 3 and 4K resolution
- Removing Friction: Automating Nano Banana Image Workflows (Dec 2025) — automated JPEG output
Project
nano-banana-experiments - View on GitHub
This post is part of my AI journey blog at Mosaic Mesh AI. Building in public, learning in public, sharing the messy middle of AI development.