// Make images & video
FLUX 3 Image Klein
A 5.6B open-weights model from Black Forest Labs (2026-10-13): text-to-image, single-image edits and compositions from up to 10 references in one model, with bounding boxes on a 0–1000 grid in the prompt and in-image text; Qwen3-VL-4B text encoder. Needs a prompt upsampler (BFL’s 9B flux-3-image-klein-pup or your own LLM) or output is generic. Peak VRAM with the reference code on B200 at 1024²: ~31 GB BF16, 26 GB FP8, 25 GB NVFP4 (upsampler excluded). FLUX Kommunity License: non-commercial, plus commercial outputs for qualifying users under US$5M revenue
Open source Maintained For everyone
IN ONE PARAGRAPH
FLUX 3 Image Klein is A 5.6B open-weights model from Black Forest Labs (2026-10-13): text-to-image, single-image edits and compositions from up to 10 references in one model, with bounding boxes on a 0–1000 grid in the prompt and in-image text; Qwen3-VL-4B text encoder. Needs a prompt upsampler (BFL’s 9B flux-3-image-klein-pup or your own LLM) or output is generic. Peak VRAM with the reference code on B200 at 1024²: ~31 GB BF16, 26 GB FP8, 25 GB NVFP4 (upsampler excluded). FLUX Kommunity License: non-commercial, plus commercial outputs for qualifying users under US$5M revenue.
Good for: Anyone who wants AI to make images & video without setting anything up.
01How to start
- 1Open the official site or app store page (link above) and sign up.
- 2Use it in the browser or download the official app.
- 3Pricing: Open source — start with the free tier if there is one.
02Alternatives & comparison
Compare side by side →Versions, why it is listed, for AI to read
For AI agents
- Interfaces：its own app or website
- Machine-readable record：/api/v1/software/flux-3-image-klein.json · llms.txt
- Added 2026-10-22 · updated 2026-10-22