Our new model for generating and editing images, trained on 100% licensed data. Same API, nothing to change in your code.
🏆 #1 open-weights image editing model on Artificial Analysis
Every result on this page was made by Fibo 2.
Text in, image out, with the VGL behind it returned on every result. Each image below is shown with what made it.
Prompt: "An early-2000s fashion editorial photograph in the desert at dusk, shot on a digital camera with a wide-angle lens from a low angle, lit with a strong off-camera flash against the ambient sky." The full prompt is in the API reference.
Prompt and a reference image of two green cocktails: "Two hands toasting with tall glasses of pink lemonade at a rooftop party at sunset, string lights in the background, playful and colorful"
Refine: the first image's VGL with the tank top changed to hot pink, same seed
curl -X POST https://engine.prod.bria-api.com/v2/image/generate \
-H "api_token: $BRIA_API_TOKEN" -H "Content-Type: application/json" \
-d '{"prompt": "A panoramic view of a neon-lit Tokyo street at night in the rain, glowing pink and cyan signs reflected in the wet asphalt, people crossing with clear umbrellas", "aspect_ratio": "21:9"}'Open Fibo 2 Image Generate in the API reference →
Say what to change. Fibo 2 changes that and keeps the rest of the image.
Combine up to four images in one edit, numbered in the order you send them. Here, a boulder, a bottle and a wordmark become one perfume poster.
"luxury fragrance poster. A single large grey boulder sits in perfectly calm, mirror-like water…" The full instruction is in the curl below.
curl -X POST https://engine.prod.bria-api.com/v2/image/edit \
-H "api_token: $BRIA_API_TOKEN" -H "Content-Type: application/json" \
-d '{
"images": [
"https://labs-assets.bria.ai/sandbox-example-inputs/fibo2_edit_still_rock.jpg",
"https://labs-assets.bria.ai/sandbox-example-inputs/fibo2_edit_still_bottle.jpg",
"https://labs-assets.bria.ai/sandbox-example-inputs/fibo2_edit_still_logo.jpg"
],
"instruction": "luxury fragrance poster. A single large grey boulder sits in perfectly calm, mirror-like water, its weathered, slightly rough stone surface catching soft light. The sky is dense, soft fog and low cloud with a warm peach-cream glow breaking through near the top, fading into cool blue-grey mist toward the horizon. The rock and its full reflection sit in the lower half of the frame.\n\nStanding upright, centered on the flat top of the boulder, is a perfume bottle. It is a square clear glass flacon with thick beveled walls and crisp edges, filled with rose-gold liquid, with a thin dip tube visible inside, a short dark metal collar and a wide matte black cap. It has no label or branding. The bottle is about one third the width of the rock. Soft diffused fog light wraps the glass, with gentle highlights on the edges, a subtle warm glow in the liquid near the base, and a soft contact shadow where it meets the stone. The bottle'\''s inverted reflection appears in the still water directly below the rock'\''s reflection, slightly darker and softer.\n\nIn the upper sky, horizontally centered about 15–20% down from the top, is a minimal logo: the word \"STILL\" in thin, widely spaced elegant serif capitals with subtle flared strokes. Beneath it, in much smaller matching capitals with the same wide spacing, is \"EAU DE PARFUM\". The text is a muted warm charcoal grey, flat and crisp, with no shadow or glow, sitting softly within the fog.\n\nMinimal, serene, high-end editorial perfume campaign. Muted palette of fog grey, slate and blue-grey, with warm peach light in the sky and the rose-gold perfume as the only warm accent. Photorealistic, shallow atmospheric depth, consistent soft lighting and perspective, fine detail, no other elements or text.",
"aspect_ratio": "1:1",
"seed": 202
}'One image works too: say what to change and Fibo 2 keeps the rest.
Before
After: "Change the car's paint to glossy cherry red, keep everything else the same"
Open Fibo 2 Image Edit in the API reference →
Fibo 2 adds 12 ratios to the nine you know, for generation and for edits with two or more reference images. Pass any of them as aspect_ratio.
"A panoramic view of a neon-lit Tokyo street at night in the rain, glowing pink and cyan signs reflected in the wet asphalt, people crossing with clear umbrellas", at 21:9
16:10 10:16 2:1 1:2 21:9 9:21 3:1 1:3 4:1 1:4 5:1 1:5
On Artificial Analysis' image editing arena, Fibo 2 is the #1 open-weights model, and it edits better than closed models from Ideogram, ByteDance and Black Forest Labs. It is also a top-2 open-weights text-to-image model.
