Tasks and inputs
FLUX 3 Image is listed for Image-text-to-image, Image-to-image, Text → Image. It accepts text, image and produces image. Use the publisher’s input format and preprocessing for your chosen deployment.
SyncDev model directory · Profile updated Nov 17, 2026
FLUX 3 Image is Black Forest Labs' single-endpoint model for image generation and editing, with ten references, bounding boxes and web grounding.
The specification score summarizes recorded features and limits. It is not a benchmark result.
Pricing conditions: Per image at launch: 768sq $0.041, 1k $0.048, 2k $0.100, 4k $0.607; custom sizes bill at the smallest tier with at least as many pixels, above ~4MP at the 4k rate. A 1.5k tier may exist. Indicative; confirm on your BFL account.
Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.
Accepts
Produces
There is no mode switch in FLUX 3 Image. One endpoint, POST /v1/flux-3-image, handles both creating a picture from nothing and editing an existing one, and the prompt alone decides which. Black Forest Labs launched it on its API on 1 October 2026. For teams that maintain separate generation and editing pipelines, that single contract is the main reason to look.
Up to ten reference images. You refer to each by its position in the prompt and say what it contributes: the scene, a product, a face, a texture.
Bounding boxes. Append coordinates to the end of the prompt as [top, left, bottom, right] on a 0-1000 grid to fix where things go. The same boxes target local edits, so you can recolour, move, replace or remove one element while the rest of the image stays put.
Web grounding. On by default. The model looks things up on the web so real-world subjects render correctly. Set grounding=false for faster results that rely on the prompt only, which is usually what you want for pure fiction or abstract art.
At launch, output ran from 768 square up to 4k (about 16MP) across 15 aspect ratios. A 22 October update added exact width and height (256 px per side up to 16.8MP, aspect ratios as extreme as 8:1, and 4MP when references are used) plus resolution=match_input, and made single-reference edits more faithful. Prices did not change.
| Tier | Price per image at launch |
|---|---|
| 768sq | $0.041 |
| 1k | $0.048 |
| 2k | $0.100 |
| 4k | $0.607 |
Custom sizes bill at the smallest tier with at least as many pixels; anything above roughly 4MP bills at the 4k rate. That jump from 2k to 4k is six-fold, so pick sizes deliberately.
The launch price table has no 1.5k tier, yet some aggregator listings show a 1.5K override at about $0.07, and the current pricing page now has a 1.5k row whose values we could not read. Request one 1.5k image and check the billed amount on your dashboard before you budget around it. All prices here are indicative.
FLUX 3 Image Klein uses the same request format at a fraction of the price and with downloadable weights. Moving a request across means changing the endpoint path. Keep the full model for photoreal hero images, complex multi-reference compositions and anything where grounding matters; send bulk edits and drafts to Klein. Other comparisons worth running are GPT-Image-2 and Gemini Nano Banana 2.1.
The API is proprietary with no weights. See Black Forest Labs on Hugging Face for the open releases, or browse text-to-image models.
Compare the model on the work you need to complete.
FLUX 3 Image is listed for Image-text-to-image, Image-to-image, Text → Image. It accepts text, image and produces image. Use the publisher’s input format and preprocessing for your chosen deployment.
Choose the output size and quality your design needs. Include drafts and revisions when budgeting an accepted image.
Use the model identifier and account requirements for the selected API route. Confirm its request limits and output format with one complete task before connecting a larger workflow.
These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.
ByteDance Seed
Seedream 5.5 is ByteDance Seed's image generation and editing model with stable 100-round editing, native RGBA output and sharper dense text.
ideogram
Ideogram Vector 1.0 generates editable SVG artwork from a prompt or converts logos and illustrations to SVG, from $0.10 per file via the API.
ByteDance Seed
Seedream 5.0 Flash is ByteDance Seed's fast, low-cost image model for generation and editing at 1K–2K, built for ad assets and posters at volume.
recraft
Recraft V4.1 Flash was Recraft's first speed-tier 1K raster image model at $0.007 per image, now superseded by Recraft V4.1 Flash V2.