![FLUX.2 [dev] logo](https://zdxtpmahqrmmxfbzzdky.supabase.co/storage/v1/object/public/content-images/logos/flux-2-dev.webp)
FLUX.2 [dev]
32B open-weight image model: multi-reference editing with local deployment.
About
FLUX.2 [dev] is the open-weight member of Black Forest Labs' November 2025 FLUX.2 family: a 32B guidance-distilled text-to-image and image-editing model that supports multi-reference editing with up to 10 reference images — the best character/product/style consistency in the open-weight space at launch — plus 4-megapixel output and reworked text rendering for typography-heavy work like UI mockups and infographics. The [dev] weights landed on Hugging Face under a non-commercial license, with FP8 and ComfyUI streaming optimizations developed with NVIDIA cutting VRAM needs by ~40%.
The honest picture: full-precision [dev] normally wants ~90GB of VRAM, so 'local deployment' realistically means quantized builds on high-end consumer GPUs. The license is research/non-commercial, so businesses need the Pro API instead. BFL later released the faster [klein] family (4B under Apache 2.0) for real-time and edge use, making [dev] the quality-over-speed choice.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Verified via vendor site/docs/GitHub and reputable coverage, wave 46; Trending score 80/100 from researcher assessment.