FLUX AI: Models, Pricing and Where to Use Them

No. Black Forest Labs develops FLUX models; Flux1.ai sells access through its own service. Flux1.ai explicitly says it is not affiliated with, endorsed by or sponsored by BFL. Its OpenTools profile covers that service, including its subscription plans. The Black Forest Labs company page connects the model developer to its products.

Choose the direct BFL Playground if you want a browser route from the developer; it requires sign-in. Choose fal.ai when a hosted API catalog fits your application. Neither a shared model name nor an alternative-tool link means the companies share ownership.

For image generation and editing, BFL documents FLUX.2 max, pro, flex, klein and dev variants. Compare the input controls and availability for the exact variant rather than treating FLUX.2 as one interchangeable endpoint. BFL also offers FLUX 3 video; this guide focuses on choosing an image workflow.

FLUX1.1 pro is an earlier text-to-image model with a distinct price and endpoint. Ultra is a separate variant: do not apply its output modes or price to standard 1.1 pro. For local generation and multi-reference editing, FLUX.2 klein 4B is a compact, distilled checkpoint with downloadable weights. Its model card reports approximately 13GB VRAM; actual requirements depend on the runtime and configuration.

Prices checked September 12, 2026, in USD. BFL lists FLUX1.1 pro at $0.04 per image and Ultra/Raw at $0.06 per image. On fal, the standard FLUX1.1 pro endpoint lists $0.04 per megapixel, rounded up. A 1.5MP image therefore bills as 2MP on that fal endpoint: $0.08. Do not assume the BFL and fal prices buy the same dimensions or settings.

For a simple planning example, 100 billable images at BFL’s $0.04 rate cost $4. That arithmetic excludes additional generations, other model variants and application costs. fal’s FLUX.2 dev endpoint displays $0.012 per megapixel; use its current endpoint price and input settings for an estimate. Hosted access to dev on fal is separate from downloading dev weights.

Flux1.ai uses platform credits. Month-to-month plans show Basic $16.90 / 300 credits, Pro $26.90 / 800, Max $35.90 / 2,000, and Pro Max $86.90 / 6,000. Annual billing totals are $142.80, $226.80, $310.80 and $718.80, respectively, with monthly credit allocations. The lower displayed monthly equivalents require the annual payment.

Credits are not images. Flux1.ai lists standard image generation at roughly 5–14 credits depending on model, so 300 credits would cover about 21–60 such generations if each used those rates. Fast Mode and other models have different debits. Subscription credits refresh monthly; separately purchased one-time credits are advertised as non-expiring. Check the selected model’s debit before a batch.

Free access and commercial permission are separate questions. Flux1.ai advertises registration credits; confirm the allowance in your account rather than treating that offer as unlimited generation. Its API requires administrator approval plus a completed paid credit-pack or subscription purchase. Free or gifted credits alone do not qualify.

For self-hosting, the official FLUX.2 klein 4B checkpoint is released under Apache 2.0. The klein 9B and dev variants have different weight licenses; do not transfer the 4B license to them. Running weights still requires compute and deployment work. BFL says its paid API includes commercial usage under the applicable terms; model training on generated output can require a separate license.

Flux1.ai’s paid plans list commercial use and private generation. Those labels do not establish a retention period or training-data policy. Check the chosen service’s terms and permissions for uploaded references, especially for client material. A model license also does not supply rights to someone else’s photo, brand or likeness.

The official FLUX MCP is another direct BFL access route. It supports image generation, reference-image edits, variations and generation history from compatible clients. Its documentation also covers video. Authentication uses OAuth with a BFL account, and the organization selected during sign-in is billed for generated media.

Choose MCP for creative work inside an assistant; choose an API for application code and a controlled production pipeline. MCP does not supply a separate free credit pool. Review what the client sends to BFL and confirm the billing organization before generating.

Start with a representative deliverable: a product concept, an illustration with readable text, or an edit that must preserve a reference. Compare the same prompt, inputs and target dimensions across the routes you are considering. Inspect typography, artifacts, reference fidelity, download quality and the actual charge. These are suggested checks, not results from a hands-on benchmark.

For a browser workflow, compare Flux1.ai’s plans with the direct BFL option. For development, start with the fal.ai profile and the exact model endpoint. For local deployment, read the klein 4B model card and limitations. Images can contain distorted text or miss the prompt; review outputs before publishing.