Flux 2 Flex vs Flux 2 Pro on Sume: $0.0625 vs $0.0375 per image
Sume's catalog lists Flux 2 Pro and Flux 2 Flex with the same 13 ratios, edit support and 4-image cap. The difference is price: 1.67x. How to test it.

On Sume, Flux 2 Pro and Flux 2 Flex differ in price and nothing else that the catalog publishes. Pro bills $0.0375 per image and Flex $0.0625, which is 1.67 times more for Flex. Both list 13 aspect ratios, take reference images, return png, jpeg or webp, and cap n at 4.
What the catalog says
| Field | Flux 2 Pro | Flux 2 Flex |
|---|---|---|
| API id | black-forest-labs/flux.2-pro | black-forest-labs/flux.2-flex |
| List per image | $0.03 | $0.05 |
| Billed (list x 1.25) | $0.0375 | $0.0625 |
| Reference images (edit) | Yes | Yes |
| Max n per call | 4 | 4 |
| Aspect ratios | 13, including 21:9, 9:21, 1:2, 2:1 | The same 13 |
| Custom pixels (image_size) | Yes | Yes |
Reading the table
Because the descriptors match, you cannot choose between them from the API shape. The choice is about the image you get for your own prompts. Sume's docs do not rank them, so this post does not either; it gives you the cost of finding out.
A 30-prompt A/B at one image per prompt costs 30 x $0.0375 = $1.125 on Pro and 30 x $0.0625 = $1.875 on Flex. Running both costs $3.00 in total.
A fair test
- Use the same prompts, the same
aspect_ratioandn: 1on both ids, so that price is the only variable you did not choose. - Include at least one edit case with an
input_referencesphoto, since both ids accept references. - Judge at delivery size. A 16:9 thumbnail is read at a few hundred pixels, so zoomed-in flaws matter less than composition.
- Log
usage.costfrom each response so that the billed amount, not an estimate, enters your sheet.
When the dearer one pays off
If Flex saves a retry, it is cheaper. With a $0.0375 Pro price and $0.0625 Flex price, one extra Pro retry per good image (two calls, $0.075) already costs more than one Flex call. If Pro lands a usable image on the first try for most prompts, stay on Pro. That is a measurement for your prompts, not a property of the models.
For ultra-wide work, both list 21:9 and 9:21, so a cinematic title card or a tall story strip does not need a different model.
Cost of a routing rule
A common setup is Pro by default and Flex for the prompts that fail review. If one in five prompts needs a Flex re-run, a batch of 100 prompts costs 100 x $0.0375 + 20 x $0.0625 = $5.00 billed. A batch run entirely on Flex costs $6.25, and one entirely on Pro costs $3.75. Your own rate of failures sets the real number, so count them for a week before you pick a rule.
Keep both ids behind a single setting in your code, and store the id that made each image next to the file. When a client asks why one image looks different from another, the answer is in that record, and a re-run on the same id is a one-line change.
Pin the id
Pin the full org/slug id and read the price from GET /v1/images/models/{id}/endpoints before large batches. The pricing line there is the amount charged to your wallet, margin included. See the Image API page for the request fields.
Sources
Related posts
More in Models
- FLUX 3 Image 5456x3072 at 300 dpi: print size vs Sume pixel caps
BFL shows FLUX 3 Image at 5456x3072 (16.8 MP). At 300 dpi that prints 18.19 x 10.24 in. Here is what Sume models give at the same dpi.
- gemini-2.5-flash-image: Google pages give Oct 2, 2026 and Mar 15, 2027
Google's pricing page says gemini-2.5-flash-image shuts down Oct 2, 2026; its deprecations page says Mar 15, 2027. How to plan around two dates.
- gemini-3.1-flash-image is deprecated: move to Nano Banana 2.1
Google deprecated gemini-3.1-flash-image on Oct 6 with no shutdown date and named gemini-nano-banana-2.1 as the replacement. What to change, and what Sume does.
- Gemini TTS 24/16/8 kHz vs Sume TTS six sample rates up to 48 kHz
Gemini 3.8 Flash TTS offers 24000, 16000 and 8000 Hz. Sume TTS 1.0 accepts six rates from 8000 to 48000 Hz. Which rate suits video, phone and web.
Written by Sume