Image Generation API Pricing Compared (2026)
byesu prices below are the live prices on 2026-10-22. Vendor prices are quoted from OpenAI pricing, the OpenAI image generation guide and Gemini API pricing (last updated 2026-10-19), checked on 2026-10-22. Vendors change prices, so confirm on their pages before budgeting.
Image models on byesu are billed per image, at one fixed price — no token math, and the price does not change between 1K, 2K and 4K. Five image models are on sale: gemini-3.1-flash-image-ultra at $0.007, gemini-nano-banana-2.1-ultra at $0.008, and gpt-image-2, gpt-image-2.5 and gemini-3.0-pro-image-ultra at $0.009 per image. All of them work with one sk- token from the media / media-gen group, pay-as-you-go with no subscription.
Image Model Pricing
| byesu model | Vendor model | byesu price (1K / 2K / 4K) | Vendor list price | Source |
|---|---|---|---|---|
gpt-image-2 | OpenAI GPT Image 2 | $0.009 / image | Token-based: $5/M text in · $8/M image in · $30/M image out (≈ $0.006–$0.211 per 1024×1024, see below) | OpenAI |
gpt-image-2.5 | OpenAI GPT Image 2.5 | $0.009 / image | Token-based, same rates as GPT Image 2; no per-image table published | OpenAI |
gemini-3.0-pro-image-ultra | Gemini 3 Pro Image (Nano Banana Pro) | $0.009 / image | $0.134 (1K / 2K) · $0.24 (4K) | |
gemini-nano-banana-2.1-ultra | Gemini Nano Banana 2.1 | $0.008 / image | $0.0336 (1K) · $0.0504 (2K) · $0.113 (4K) | |
gemini-3.1-flash-image-ultra | Gemini 3.1 Flash Image (Nano Banana 2) | $0.007 / image | $0.067 (1K) · $0.101 (2K) · $0.151 (4K) |
Google's figures are its standard-tier per-image equivalents of image output tokens; its Batch tier is about half. Live prices for every model are also shown in the Generation Playground.
For scale: 100 images at 4K with Nano Banana Pro come to about $24.00 at Google's standard list price and $0.90 with gemini-3.0-pro-image-ultra on byesu.
GPT Image 2 and 2.5: From Token Pricing to a Per-Image Cost
OpenAI does not sell its image models by the image. GPT Image 2 is billed by token: $5 per million text input tokens, $8 per million image input tokens and $30 per million image output tokens. Output tokens grow with size and quality, so the cost moves with every request. OpenAI's own examples for GPT Image 2:
| Quality | 1024×1024 | 1024×1536 / 1536×1024 |
|---|---|---|
| Low | ≈ $0.006 | ≈ $0.005 |
| Medium | ≈ $0.053 | ≈ $0.041 |
| High | ≈ $0.211 | ≈ $0.165 |
Edit requests add image input tokens for every reference image, and 2K or 4K sizes cost more than the sizes above. OpenAI sells GPT Image 2.5 as two model IDs (gpt-image-2.5-sunburst and gpt-image-2.5-flare) at the same token rates, without a per-image table.
One honest note: for a small square image at low quality, calling OpenAI directly (≈ $0.006) is cheaper than byesu's $0.009. The flat price pays off at medium and high quality and at 2K/4K sizes, where OpenAI's token bill rises and byesu's stays at $0.009.
Which Model Should You Pick?
| Model | Good for | byesu price |
|---|---|---|
gemini-3.1-flash-image-ultra | Fast drafts and high-volume batches (Google positions Nano Banana 2 for speed and throughput) | $0.007 — lowest |
gemini-nano-banana-2.1-ultra | Google's update to Nano Banana 2: better visual quality, character consistency across turns, text rendering | $0.008 |
gemini-3.0-pro-image-ultra | Nano Banana Pro, Google's top image tier; the biggest gap to the vendor list price | $0.009 |
gpt-image-2 | Instruction-faithful everyday images; edits with multiple reference images | $0.009 |
gpt-image-2.5 | OpenAI's newer generation — try it when gpt-image-2 misses detail | $0.009 |
Rule of thumb: iterate on gemini-3.1-flash-image-ultra, then render finals on the model whose look you prefer. Switching is a one-string change; the token stays the same.
Video Models
No video model is on sale as of 2026-10-22, so this page compares image models only. When video returns, it will be billed per second (unit price × seconds), failed tasks will be refunded automatically, and the models and prices will be listed on Video Generation.
How to Call These Models
gpt-image-2/gpt-image-2.5: OpenAI-compatiblePOST /v1/images/generations(text-to-image) andPOST /v1/images/edits(reference images).- The three Gemini models:
POST /v1beta/models/{model}:generateContent, withgenerationConfig.imageConfig.imageSizeset to"1K","2K"or"4K".
Use a token from the media / media-gen group — a token from another group returns "no available channel". Details: Image Generation, Calling via API and Choosing a group. Top up with USDT or Alipay; there is no subscription or minimum spend.
FAQ
Which image generation API is the cheapest on byesu?
gemini-3.1-flash-image-ultra (Nano Banana 2) is the lowest at $0.007 per image, followed by gemini-nano-banana-2.1-ultra at $0.008. gpt-image-2, gpt-image-2.5 and gemini-3.0-pro-image-ultra (Nano Banana Pro) are $0.009 per image. Each price is the same at 1K, 2K and 4K. Prices checked on 2026-10-22.
How much does the gpt-image-2 API cost per image?
On byesu, gpt-image-2 is $0.009 per image at any resolution. OpenAI bills the model by token ($30 per million image output tokens); its own examples for a 1024×1024 image are about $0.006 at low quality, $0.053 at medium and $0.211 at high quality (OpenAI image generation guide, checked 2026-10-22).
What does Nano Banana Pro cost per image?
Google lists Gemini 3 Pro Image (Nano Banana Pro) at about $0.134 per 1K or 2K image and $0.24 per 4K image on the standard tier (Gemini API pricing page, last updated 2026-10-19). On byesu, gemini-3.0-pro-image-ultra is $0.009 per image at 1K, 2K or 4K.
Does a 4K image cost more than a 1K image?
Not on byesu: all five image models on sale charge the same per-image price at 1K, 2K and 4K. At the vendors, price rises with resolution — for example, Gemini 3.1 Flash Image is listed at $0.067 for 1K and $0.151 for 4K.
Is a video generation API available on byesu right now?
No. As of 2026-10-22 no video model is on sale, so this page compares image models only. When video returns it will be billed per second with automatic refunds for failed tasks, and the Video Generation page will list the models and prices.
Related Links
- Image Generation — models, resolutions and reference images
- Calling via API — request formats for both endpoint styles
- Model list — every model currently on sale
- byesu vs OpenRouter — how the two gateways bill