Cheapest AI image API per image in October 2026: vendor rates
Lowest listed per-image prices on Google, Black Forest Labs, fal and Runway pages read today, and why a sticker price is not your cost per image on Sume.

On the vendor pages read today, the lowest sticker prices are Runway's Muse Image at 1 credit (about $0.01), FLUX.2 [flex] at $0.024 per image on Black Forest Labs, and Gemini 3.1 Flash Lite Image at $0.0336 for a 1K image. They are not comparable one to one, because each page bills a different unit, so the honest answer is a short ladder rather than a single winner.
Below is what each page said on 2026-10-02, followed by how to turn a sticker price into a number you can put in a budget on Sume.
What do the vendor pages list per image?
Every number here comes from the vendor's own page. Where a page gives a range, the range is quoted, not a single point.
| Vendor page | Model | Listed price | What the page says about the unit |
|---|---|---|---|
| Runway API pricing | Muse Image | 1 credit | Credits cost $0.01 each in the developer portal |
| Black Forest Labs pricing | FLUX.2 [flex] | $0.024 per image | Calculator; price scales with resolution (768x768 up to 4K) |
| Black Forest Labs pricing | FLUX.2 [pro] | $0.048 per image | Same calculator, resolution-dependent |
| Gemini API pricing | Gemini 3.1 Flash Lite Image | $0.0336 per 1K image | Token-billed, per-image figure shown for 1K |
| Gemini API pricing | Gemini 3.1 Flash Image | $0.045 (512px), $0.067 (1K) | Token-billed; 2K is $0.101 and 4K is $0.151 |
| fal pricing | Nano Banana to GPT Image 2 | $0.0398 to $1.00 per unit | Resolution, duration and quality can change the final cost |
Why is the cheapest sticker price not the cheapest image?
Three things move the real cost. First, the unit: BFL's calculator scales by resolution, Google bills output tokens, and fal quotes a per-unit range that spans a 25x gap between its cheapest and dearest image models. Second, the resolution you actually need: a 512px draft and a 4K final are different products. Third, retries: the cost per keeper is the price divided by your keep rate, so a model that is half the price but needs twice the attempts costs the same.
A practical rule is to price the same prompt at the same output size on every candidate, then divide by the number of images you keep.
How do I get a real per-image number on Sume?
Sume bills a single USD wallet. For listed models, the per-image line comes from the endpoint record, and the image docs state that "Endpoint pricing lines are the amount charged to your wallet", with Sume's margin already applied, so cost_usd × n is what you pay. Read the line before you commit to a batch.
Image generation billing is all-or-nothing: a completed generation is billed in full, and a failed or cancelled one is not billed. That removes failed attempts from the cost-per-keeper math, though not the images you generate and then discard.
curl "https://api.sume.com/v1/images/models" \
-H "Authorization: Bearer $SUME_API_KEY"
curl "https://api.sume.com/v1/images/models/bytedance-seed/seedream-4.5/endpoints" \
-H "Authorization: Bearer $SUME_API_KEY"
# read endpoints[].pricing[].cost_usd, then multiply by nWhat does Sume not do here?
Sume does not match another vendor's sticker price, and it does not publish a single cheapest image model: the answer depends on the model you pick and the resolution tier you request. The rule for listed SKUs is provider list price times 1.25, reserved from the wallet when you submit and settled when the job finishes. See Generation admission for the reserve and refund behavior.
If you are comparing, use the same size and quality on both sides, and compare the Sume endpoint line against the vendor's per-image figure for that exact tier. A Runway credit is $0.01, so Muse Image at 1 credit is a useful floor for the comparison, but check that the model is the one you want before you let price decide.
Sources
Related posts
More in Pricing
- Cheapest speech-to-text API per hour, October 2026
Per-hour STT cost from xAI, OpenAI and Google pages against Sume's $0.01 per minute, with a candid note on where Sume is the expensive option.
- Cheapest TTS API per 1,000 characters, October 2026
List prices per 1,000 characters from Mistral, xAI, Speechify, Artificial Analysis and Sume, converted from each vendor's own units, with the trade-offs.
- Cost of 1,000 six-second AI video clips by model
A batch of 1,000 six-second clips costs $75 on Grok Imagine Video 1.5 and $750 on Wan 3.0 at 720p. Full Sume table plus how to submit it without queue_full.
- Cost of a 30-second AI video ad: clips, captions, music
A 30-second AI ad costs $0.80 on Grok 720p and up to $6.73 on Kling 3 audio on, once clips, timeline render, captions and music are added. Sume prices.
Written by Sume