Navigation
Explore
Explore
Flux by Black Forest Labs is an AI image generation model family for developers and designers with open weights, API access, and FLUX Kontext image editing.
Image Generation
9.5FLUX.2 model family covers real-time generation at $0.014/image (Klein), production quality at $0.03 (Pro), and maximum quality at $0.07 (Max); accurate text rendering in generated images; FLUX.1 Kontext adds in-context image editing; open-weight variants (Schnell, Klein) enable self-hosted pipelines.
Design
9.2FLUX.1 Kontext enables text-instruction-based editing (change clothing, alter backgrounds, add/remove objects) while preserving non-edited regions — covering iterative design revision workflows without full regeneration; FLUX Fill handles mask-based inpainting for selective region editing in product and creative design contexts.
Content Creation
9.0Accurate text rendering in generated images covers marketing graphics, social media visuals, and branded content requiring readable text; LoRA fine-tuning on FLUX.2 Klein enables consistent style personalisation across content series; FLUX.1 Schnell's 4-step fast generation covers high-frequency content creation pipelines.
Automation
9.2REST API with credit-based per-image pricing enables programmatic integration into automated image generation pipelines; MCP integration available for agent-based workflows; batch request pricing scales linearly for high-volume automated workflows; sub-second FLUX.2 Klein generation supports real-time automation requirements.
Marketing
8.8High photorealism from FLUX.2 Pro and Max tiers covers advertising creative production; text rendering accuracy enables marketing graphics with readable copy without separate text overlay tools; FLUX.1 Kontext covers product background replacement and style adaptation for campaign variant generation.
Flux is a family of AI image generation and editing models developed by Black Forest Labs (BFL), co-founded by the original Stable Diffusion authors and backed by Andreessen Horowitz. The FLUX.2 generation offers megapixel-based pay-per-image API pricing from $0.014 (FLUX.2 Klein for real-time use) to $0.07 (FLUX.2 Max for highest quality). FLUX.1 Kontext enables in-context image editing via text instructions — modifying clothing, backgrounds, expressions, and objects while preserving non-edited regions. Open-weight models include FLUX.1 Schnell (Apache 2.0, commercial) and FLUX.2 Klein (Apache 2.0). FLUX.1 Dev is non-commercial licence only. BFL provides a Playground at bfl.ai and API at api.bfl.ml; consumer-facing use is primarily through third-party platforms (Replicate, fal.ai, Freepik). Training data is not publicly disclosed.
Pricing
| Plan | Model | Usage Limits | Price |
|---|---|---|---|
| BFL API | FLUX.2 [klein] 4B and 9B (sub-second generation); FLUX.2 [pro] (production quality, fast); FLUX.2 [max] (highest quality, grounding search); FLUX.2 [flex] (multi-reference, hex colour control); FLUX1.1 [pro] and [pro] Ultra/Raw; FLUX.1 Kontext [pro] and [max] (image editing); FLUX.1 Fill [pro] (inpainting/outpainting) | 1 credit = $0.01 USD; no monthly subscription; credits purchased via dashboard; same pricing for API and Playground; FLUX.2 Klein from $0.014/image; FLUX.2 Pro from $0.03 text-to-image / $0.045 editing; FLUX.2 Max from $0.07/image; FLUX.2 Flex $0.06/image; FLUX1.1 Pro $0.04/image; FLUX.1 Kontext [pro] from $0.04/image; FLUX.1 Kontext [max] higher pricing; FLUX.2 pricing scales with output resolution (megapixel-based); batch requests multiply base cost by image count; rate limits not publicly documented per tier | pay-as-you-go, credit-based |
| FLUX.1 [dev] | 12B parameter model; HuggingFace download; non-commercial licence; full diffusers integration | — | open weights, non-commercial |
| FLUX.1 [schnell] | 12B parameters; 4-step fast generation; commercial use permitted; Replicate ($0.003/image) and HuggingFace available | — | open weights, Apache 2.0 |
| FLUX.2 [klein] | Sub-second generation; commercial use permitted; LoRA fine-tuning and style training supported | — | open weights, Apache 2.0 |
REST API with credit-based PAYG pricing from $0.014/image eliminates subscription overhead for variable-volume applications; open-weight Apache 2.0 models (Schnell, FLUX.2 Klein) enable self-hosted deployment without API costs; MCP and Agent Skills integrations cover AI agent workflows; batch pricing scales linearly with volume for high-throughput pipelines.
FLUX.1 Kontext in-context editing handles iterative design changes (clothing, backgrounds, expressions, objects) via text instructions without full regeneration, reducing design iteration time; FLUX Fill handles mask-based inpainting for selective region editing; LoRA fine-tuning on FLUX.2 Klein creates consistent branded style models for repeatable design outputs.
FLUX.1 Schnell's 4-step fast generation and FLUX.2 Klein's sub-second generation cover high-frequency social media content creation pipelines at low per-image cost; accurate text rendering in generated images reduces the need for separate text overlay in social graphics; third-party platforms (Freepik, fal.ai) provide consumer-friendly UX on top of BFL models for creators without API access.
Credit-based API pricing with no seat costs or subscriptions enables agencies to pass per-image generation costs directly to client projects without fixed overhead; FLUX.1 Kontext product background replacement and variant generation covers campaign creative production; LoRA fine-tuning enables per-client style models for consistent brand visual identity; FLUX.2 Max at $0.07/image covers premium agency deliverables requiring highest quality output.
Midjourney
Consumer-first UX with curated aesthetic house style; $10/month subscription; no API or open weights; stronger for non-technical creators prioritising aesthetic output; no image editing via text instructions.
Compare →
DALL-E 3
Integrated into ChatGPT; simple safety filter; no open weights or API-only access; $0.040/image via OpenAI API; weaker resolution and complex scene fidelity than FLUX.2 Pro or Max.
Compare →
Adobe Firefly
IP-cleared training data with Content Credentials; Adobe Creative Cloud integration and Photoshop Generative Fill; Firefly credits bundled with CC; weaker raw image quality than FLUX.2 Max; stronger enterprise IP compliance position.
Compare →
Stable Diffusion
Compare →
Consider These Instead
Choose Midjourney when aesthetic coherence, curated style presets, and community-driven creative workflows for non-technical users are the priority — Midjourney's house style and guided generation UX outperform Flux for consumer creative use cases without API configuration. Choose Adobe Firefly when IP-safe generation with commercially cleared training data, Adobe Creative Cloud integration, and Photoshop/Illustrator Generative Fill are required for commercial design workflows — Firefly's Content Credentials and cleared dataset address the IP provenance concerns that Flux's undisclosed training data creates. Choose Stable Diffusion (via Automatic1111 or ComfyUI) when the largest community ecosystem, most extensive LoRA/extension library, and deepest local pipeline customisation are required for research or artistic workflows — the SD ecosystem's community depth exceeds BFL's self-hosted option.
