Nano Banana 2.1 vs GPT Image 2.5: 4.04x at 1K, 0.9x at 4K
On Sume, Nano Banana 2.1 costs $0.10 at 1K and $0.20 at 4K; GPT Image 2.5 costs $0.02475 low and $0.2225 high. The ratio flips between 2K and 4K.

On Sume, Nano Banana 2.1 is $0.10 per 1K image and GPT Image 2.5 on low quality is $0.02475, so Nano Banana costs 4.04 times as much at the bottom. At 4K the order reverses: Nano Banana 2.1 is $0.20 and GPT Image 2.5 on high is $0.2225, so GPT Image 2.5 costs 1.11 times as much. The two Sume ids are nano-banana-2.1 and gpt-image-2.5 on the Image Router.
The three matched points
GPT Image 2.5 is tiered by quality and size rather than by one resolution field, so the matched points below follow the Sume catalog rows: low at 1K, medium at 2K and high at 4K. Nano Banana 2.1 is priced per tier.
| Size | Nano Banana 2.1 | GPT Image 2.5 (quality) | Ratio | Arithmetic |
|---|---|---|---|---|
| 1K | $0.10 | $0.02475 (low) | 4.04 | 0.10 / 0.02475 |
| 2K | $0.15 | $0.055625 (medium) | 2.70 | 0.15 / 0.055625 |
| 4K | $0.20 | $0.2225 (high) | 0.90 | 0.20 / 0.2225 |
What the ratio means for a batch
For 200 images, the 1K gap is $20.00 against $4.95. At 2K it is $30.00 against $11.125. At 4K, Nano Banana 2.1 is $40.00 and GPT Image 2.5 on high is $44.50. GPT Image 2.5 wins on price at 1K and 2K, and Nano Banana 2.1 wins at 4K.
The quality tiers are not interchangeable. A low-quality GPT Image 2.5 draft and a high-quality final are different outputs, and this post makes no claim about which model looks better. Dense text and packaging are the cases where Sume's own docs suggest using a higher quality value.
Calling either one
Both ids go through the Images API with the same job envelope. Nano Banana 2.1 takes a resolution tier, and GPT Image 2.5 takes a quality value. The docs also note that the retired nano-banana-2 id still works and runs as 2.1.
curl -X POST https://api.sume.com/v1/images \
-H "Authorization: Bearer $SUME_API_KEY" \
-H "Content-Type: application/json" \
-H "Idempotency-Key: nb21-4k-001" \
-d '{
"model": "google/nano-banana-2.1",
"prompt": "Studio shot of a ceramic mug on a linen cloth",
"resolution": "4K",
"aspect_ratio": "4:5"
}'Rule of thumb
Draft on GPT Image 2.5 low at $0.02475, then generate finals at 4K on Nano Banana 2.1 at $0.20 if the 4K size is what you ship. Each of those two steps is a separate paid job.
Other sizes
Nano Banana 2.1 also has a 0.5K tier at $0.075. GPT Image 2.5 has no matching published row in the shared price list for 0.5K, so the grid starts at 1K. The catalog default estimate for GPT Image 2.5 is $0.2225, the high 4K figure, which is the number Sume reserves when you do not set a quality. If you leave quality at the Images API default for GPT Image 2.5, check the job estimate before a large batch.
Per-image prices are charged once per generated image, so asking for four images of one prompt multiplies every figure above by four. At 2K that is $0.60 for Nano Banana 2.1 and $0.2225 for GPT Image 2.5 medium.
Sources
Related posts
More in Comparisons
- Nano Banana 2.1 vs GPT Image 2.5 on Sume: 4.0x at 1K, 0.9x at 4K
Sume lists Nano Banana 2.1 at $0.10, $0.15 and $0.20 and GPT Image 2.5 at $0.02475, $0.055625 and $0.2225. The price ratio flips below 1 at 4K.
- One provider behind Hy Image 3.5 Preview vs Sume's one image endpoint
OpenRouter forwards Hy Image 3.5 Preview to one provider, Tencent Cloud. Sume's image API serves every model from one endpoint. What each means for routing.
- OpenRouter signs timestamp,body; Sume signs timestamp.body
The two webhook signatures differ by one separator and a header shape. Side-by-side table and a tested Python verifier for both, refusing empty secrets.
- OpenRouter's workspace default callback URL vs Sume's callback_url
OpenRouter has a workspace default for video callbacks. Sume's docs describe callback_url or webhook_url on each request. A Python submit wrapper.
Written by Sume