Black Forest Labs pricing calculator vs Sume's image model catalog
BFL's pricing page is a calculator by model, resolution and image count. Sume publishes models and pricing in GET /v1/images/models. Where each price lives.
Where do you find the price of a FLUX image? On Black Forest Labs' pricing page it is an interactive calculator: pick a model, a resolution and a number of images, and it shows a rate per image and a total. On Sume it is data: the Image API docs say you can explore available models and pricing with GET /v1/images/models, so a program can read the catalog instead of a person reading a page.
What does the BFL page show?
The pay-as-you-go section says there are no subscriptions and no seat fees, only what you generate. The calculator has Image and Video tabs. For images it offers these models: FLUX 3 Image, FLUX.2 flex, FLUX.2 pro, FLUX.2 max, FLUX.2 klein 4B, FLUX.2 klein 9B, FLUX Outpainting, FLUX Eraser and FLUX VTO. Resolution choices are 768 x 768, 1K, 1.5K, 2K and 4K, plus an image count. In the static text I could read, the rate field showed $0.048 and $0.024 per image, which are placeholder values for whatever selection the calculator is on, so this post does not assign them to a model.
The same page lists open-weights licensing tiers (Builder at 10K images a month, Platform and Professional at 100K, plus Enterprise and Synthetic Data), which are a separate product from the hosted API.
| Element | What the page lists |
|---|---|
| Billing model | Pay as you go; no subscriptions, no seat fees |
| Image models in the calculator | FLUX 3 Image; FLUX.2 flex, pro, max, klein 4B, klein 9B; Outpainting; Eraser; VTO |
| Resolution options | 768 x 768, 1K, 1.5K, 2K, 4K |
| Enterprise | Custom pricing, volume discounts, SLA, dedicated support |
What does Sume expose?
The Image API docs list three endpoints: POST /v1/images, GET /v1/images/models and GET /v1/images/models/{model_id}/endpoints. Parameters a model accepts are published as capability descriptors on the catalog, so you can check what a model supports before you call it. Video has the same idea: GET /v1/videos/models returns, for each model, supported resolutions, aspect ratios, durations and a pricing_skus object.
On a generation response, usage.cost is the billed USD amount for images, and token counts are reported as 0 in v1. For image models priced by tokens, the docs show the working: the Flare and Sunburst variants of ChatGPT Image 2.5 use the same Fal token rates, $30 per million output image tokens, $8 per million input image tokens and $5 per million input text tokens, before Sume pricing.
Why does the difference matter?
A calculator suits one-off budgeting: choose the combination and read the figure. A catalog suits a program that must choose a model per request, or a finance script that must reproduce a quote. If you want to compare FLUX with something else, read the live BFL page for the model and resolution you plan to use, then call the Sume catalog and take the price for the models it lists, and do the multiplication yourself on the day you decide.
Sources
Related posts
More in Comparisons
- Bilingual captions: continuous language detection vs a language hint
MAI-Transcribe-2-Streaming detects language continuously across 60 languages. Sume's caption `language` is a hint. What that means for bilingual clips.
- Cartesia overage $38-$65 per million credits vs Sume TTS price
Cartesia charges overage of $65, $45 or $38 per million credits by plan. Sume TTS lists $47.50 per million characters. Where each one wins, with the math.
- Claude Code mod vs MCP server vs skill vs hook: where Sume fits
Claude Code mods, MCP servers, skills and settings hooks overlap. Which one gives an agent Sume's image and video tools, and which one only guards them.
- Colossyan 20-30 minute video cap vs Sume avatar videos of 4-60 seconds
Colossyan's pricing page lists a per-video limit of 20 to 50 minutes and 40 scenes. Sume's avatar video takes 4 to 60 seconds per job.
Written by Sume