FLUX.2 Klein 4B BF16
256–1024px resolution (16px steps). Default: 1024×1024, 4 steps. Guidance: 1.0 (fixed). Unified model for generation and multi-reference editing. Sub-second inference.
Transform any image using natural-language prompts.
Estimated cost: calculating… $0.15510/image
Available only to logged in users
{
"image": "@Gpt26_img2img_src1_street_portrait_gpt26.png",
"prompt": "Turn this into a professional studio headshot: replace the street with a seamless mid-grey studio backdrop and soft key lighting. Keep his face, beard, hair, expression, the denim jacket and the white t-shirt exactly as they are.",
"model": "gpt-image-2.6",
"seed": 527999430
}
curl -s -o 'Gpt26_img2img_src1_street_portrait_gpt26.png' 'https://assets.deapi.ai/23635551/Gpt26_img2img_src1_street_portrait_gpt26.png'
curl -X POST \
'https://api.deapi.ai/api/v2/images/edits' \
-H 'Accept: application/json' \
-H 'Authorization: Bearer {{ YOUR_API_TOKEN }}' \
-F 'image=@Gpt26_img2img_src1_street_portrait_gpt26.png' \
-F 'prompt=Turn this into a professional studio headshot: replace the street with a seamless mid-grey studio backdrop and soft key lighting. Keep his face, beard, hair, expression, the denim jacket and the white t-shirt exactly as they are.' \
-F 'model=gpt-image-2.6' \
-F 'seed=527999430'
Built for developers who need power without complexity.
256–1024px resolution (16px steps). Default: 1024×1024, 4 steps. Guidance: 1.0 (fixed). Unified model for generation and multi-reference editing. Sub-second inference.
Edit an existing image with a plain-language instruction. Change one element, replace a marked region, restyle the whole frame, translate text in place or finish a sketch. Pass more than one image to combine people, products and settings from separate references.
Edit an existing image with a plain-language instruction. Swap an object, change a colour, add signage, remove people or restyle the whole frame while the subject, composition and background stay intact. Pass a second image to bring a product or a reference into the scene.
· Experimental · Partner model · Premium · Edit at high input fidelity, so detail in the source survives the round trip. Combine up to two references, restyle a subject, or change one element of a composition while the rest holds.
· Experimental · Partner model · Premium · Add, remove, combine or restyle elements of an existing image while composition, lighting and fine detail hold. Supply up to two reference images and describe the change in natural language — no masking required.
256–1024px resolution. 1–50 inference steps. Default: 768×768, 40 steps. Natural language image editing with AI-powered transformations.
Join developers & vibe coders building AI apps with deAPI. No credit card, zero setup headaches — just pure, instant AI magic.
Need enterprise pricing or SLA? Talk to Sales
You've reached your free daily limit
Your counter resets at 00:00 UTC.
Want to keep going? Register now and get $5 in free credits to start.
Quick check before we generate