Google Lists Nano Banana 2.1: $1.50 Input, 66k Context, Image+Text I/O

What Google Dropped

Google quietly listed Nano Banana 2.1 on October 6, 2026. It's a multimodal model — image+text in, image+text out — priced at $1.50 per 1M input tokens and $7.50 per 1M output tokens, with a 66k context window.

No benchmark claims, no research paper drop. Just a listing.

Does the Pricing Make Sense for Your Workload?

The input cost is reasonable for a multimodal model — $1.50/1M keeps it accessible for moderate-volume pipelines. The output side at $7.50/1M is where costs stack up fast; if your use case is generation-heavy (long captions, document synthesis, image generation), run the numbers carefully before committing.

The 66k context window sits in a practical middle range. It handles long documents and multi-turn conversations without burning tokens the way larger context windows can on chunked retrieval tasks. For most RAG setups, 66k is workable but not exceptional.

The bidirectional image+text modality is the real differentiator here. Models that accept and output images are still a smaller pool than text-only or image-input-only alternatives — worth noting if your pipeline needs generated visuals alongside text.

Worth Watching?

If you're building multimodal workflows with moderate output volume, Nano Banana 2.1 is worth a benchmark run. High-output or cost-sensitive teams should compare it directly against alternatives in the same price tier before locking in.

See full specs, context limits, and price history: Nano Banana 2.1 — live specs & price history

Ezra, Scout AI Team