nano-banana-2 still works on Sume: it runs as Nano Banana 2.1 at 10c
Nano Banana 2 is retired on Sume. The ids nano-banana-2 and google/nano-banana-2 still work and run as Nano Banana 2.1, and the job stores the 2.1 id.

Nano Banana 2 is retired on Sume, but the ids nano-banana-2 and google/nano-banana-2 still work. Sume runs them as Nano Banana 2.1 (google/nano-banana-2.1) and stores the 2.1 id on the job. The retired id is not in the model catalog, and a 2.1 image bills 10 cents.
What changes for an old integration
Nothing breaks, but two things shift under you. The model on the job is now 2.1, so a stored job's model field will read the new id. And your cost is the 2.1 rate, which the catalog gives as $0.08 list at 1K, times 1.25, equals 10 cents billed.
| Id you send | What runs | Billed per image at 1K | 1,000 images |
|---|---|---|---|
| nano-banana-2 | Nano Banana 2.1 | 10 cents | $100 |
| google/nano-banana-2 | Nano Banana 2.1 | 10 cents | $100 |
| google/nano-banana-2.1 | Nano Banana 2.1 | 10 cents | $100 |
| nano-banana-2.1 (bare id) | Nano Banana 2.1 | 10 cents | $100 |
Why to update the id anyway
Sume accepts bare Image Router ids such as gpt-image-2 and nano-banana-2.1 as aliases for their org/slug equivalents, so there is no urgency. But a pinned id in your code and logs should say what really runs. If you compare outputs across weeks, two jobs both labelled nano-banana-2 would be the same 2.1 model, and a log that says 2.1 is easier to audit.
The fix is a one-line change in each call: use google/nano-banana-2.1. Look up the live descriptor with GET /v1/images/models/google/nano-banana-2.1/endpoints to confirm ratios and price. Ratios listed on 2.1 include 4:5, 9:16, 16:9 and the long ones 4:1, 1:4, 8:1 and 1:8.
- Resolution tiers on 2.1: 0.5K, 1K, 2K, 4K, with charges that scale away from 1K.
- Up to 4 images per call; edits are supported.
- If you want the Pro row instead, change the id to Nano Banana Pro and expect 19 cents.
Search your code
Grep for nano-banana-2 in config files, prompts and stored job records, then replace the id. Keep a note of the date, since runs before and after the retirement of the old model may differ in look even with an unchanged prompt.
What to tell your team
Tell them three things. The old id still works. The model that runs is 2.1. The bill is 10 cents at 1K per image, so 1,000 images are $100 and 10,000 are $1,000. If anyone relied on the earlier model's output style, regenerate a small sample on 2.1 before the next campaign.
The retired id is not in GET /v1/images/models, so a dropdown built from the catalog will not show it. Anything that still sends it keeps working, which is exactly why it can linger in old scripts for months.
Sources
Related posts
More in Models
- Nano Banana 2 shutdown on Oct 29? Google's changelog gives no date
Some pages quote an Oct 29 shutdown for Nano Banana 2. Google's Oct 6 changelog lists no date. What that means for a Sume image call.
- Packaging label text test: 12 prompts on 3 Sume models, what it costs
Test Nano Banana 2.1, Imagen 4 Ultra and GPT Image 2.5 on 12 label-text prompts first: about $1.20, $0.90 and $1.40 on Sume. A short script runs it.
- Pin lyria-3.5 or leave sume/music-auto? Read routed_model per job
sume/music-auto resolves to Lyria 3.5 today and can move without notice. When to pin lyria-3.5 on Sume and how job.request.routed_model records the engine.
- Premium image models on Sume ranked: 7 to 27 cents a picture
ChatGPT Image 2.5 high is 7 cents on Sume, Nano Banana 2.1 and Qwen Image Max 10, Nano Banana Pro 19, ChatGPT Image 2 27. Ranked, with caveats.
Written by Sume