FLUX 3 Image is on BFL's page: which FLUX id does Sume call?
BFL lists FLUX 3 Image, FLUX.2 Max and Klein. Sume's image router calls FLUX.2 pro and flex. Ids, prices at $0.0375 and $0.0625, ratios and edit limits.

Sume's image router calls FLUX.2 [pro] and FLUX.2 [flex] today: black-forest-labs/flux.2-pro and black-forest-labs/flux.2-flex. Black Forest Labs' models page now leads with FLUX 3 Image, described as letting you place every element on a canvas and edit a finished image one box at a time, and it also lists FLUX.2 Max, FLUX.2 and FLUX.2 Klein. None of those newer names is a Sume catalog id on 2026-10-08.
If you call black-forest-labs/flux.3-image you will get 404 model_not_found, because the model field is validated against the catalog. Pick one of the two FLUX rows that exist.
What the router has
| Field | FLUX.2 pro | FLUX.2 flex |
|---|---|---|
| Sume id | black-forest-labs/flux.2-pro | black-forest-labs/flux.2-flex |
| Billed per image (list x 1.25) | $0.0375 | $0.0625 |
| Aspect ratios listed | 13, incl. 4:5, 21:9, 9:21, 2:1, 1:2 | same 13 |
| Edits with references | up to 10 | up to 10 |
| Custom pixels | image_size {width,height} | image_size {width,height} |
| Output formats | png, jpeg, webp | png, jpeg, webp |
The FLUX 3 box-edit idea on Sume
FLUX 3 Image's pitch is region control: change one box and leave the rest. Sume has no bounding-box field. The tools you do have are different: a reference image plus a prompt that names what to change, aspect_ratio: "auto" to keep the source shape, and mask_url, which only ChatGPT Image 2.5 accepts.
So a region edit on FLUX.2 is a prompt edit, and the rest of the frame can still shift. When you need the untouched area to stay identical, either use the ChatGPT Image 2.5 mask path or paste the edited region back over the original in your own code.
Money and ratios first
On price alone FLUX.2 pro at $0.0375 sits between Seedream 5.0 Lite ($0.04375) and Qwen Image ($0.025). Its edge on Sume is the ratio list: 13 values, including 21:9, 9:21, 2:1 and 1:2, which Imagen and Grok Image do not offer for 4:5, and which Seedream does not offer for the extreme crops. Flex costs two thirds more per image, so test whether the difference shows in your prompts before you switch.
Each row returns a hosted URL under media.sume.com, not base64. Download the file if you need it past the signed URL's life.
Choose a row in three questions
- Need a precise mask? Use
openai/gpt-image-2.5withmask_url. - Need 21:9 or 9:21 at about four cents? FLUX.2 pro lists both.
- Need text on a label? Compare FLUX.2 flex with Ideogram 4.5 on a ten-headline test before you commit.
What Sume does not do
Sume does not expose FLUX step counts or guidance values: provider.options must be empty or {}. It also does not add FLUX 3 until the catalog lists it. The live list is GET /v1/images/models; check it before you assume a new BFL name is callable, and read BFL's page for what the newer models do.
Sources
Related posts
More in Models
- Which new text models can a Sume Format run name as model?
Format runs take a catalog id as model. Haiku 5.5, GLM 5.3 Flash and GPT-6 Astra are in Sume's registry; Mistral Large 4 and GLM 5.3 Fast are not.
- Gemini Flash-Lite TTS lacks 29 languages that Flash covers
Google's table lists 132 languages for Gemini 3.8 Flash TTS and 103 for Flash-Lite, so 29 are Flash-only, including Thai, Swedish and Finnish. Read 2026-10-08.
- Gemini Omni Flash 1.1 at 4K vs 1080p for an 8-second product ad
An 8-second Gemini Omni Flash 1.1 clip costs $1.00 at 720p, $1.50 at 1080p and $3.00 at 4K on Sume. Audio is always on and the ratio is 16:9 or 9:16.
- Gemini Omni Flash 1.1 on Sume: spec sheet and billing in one page
Everything that ships for gemini-omni-flash-1.1 on Sume: 3 to 10 seconds, four resolutions, two ratios, always-on audio, reference limits, edit mode, billing.
Written by Sume