We had the opportunity to test Reve’s model a while back. Reve 2.0 picks up where the last model left off. It is a superb 4K image model that can understand complex prompts. It ranks #2 on Arena’s text to image model, ahead of Nano Banana 2. Each image is segmented, so you will have more editing control over each region.
Today, we’re launching Reve 2.0, the best 4K image model in the world.
We invented a new way to generate and edit any image using precise layouts. For the first time, it’s possible to create images you can touch. pic.twitter.com/mdj2xDEqfp
— Reve (@reve) June 3, 2026
Images are treated like code, so you can edit each part. We tested some of our most complex prompts and got a few stunning results. Here is one:
16:9, 2×2 grid, 4 economic systems as typographic defining symbols (Capitalism / Feudalism / Mercantilism) Anchor: `[Symbol] :: [Economic System / Historical Period (e.g., Capitalism / Feudalism / Mercantilism)] :: 4` Morphology: `3D typographic sculpture of [Symbol], form assembled from a dense pile of [Economic System] daily-life objects — the material culture produced by and sustaining that system — tools of labour, tokens of exchange, instruments of production, documents of ownership — chaotic, legible :: 3` Material Physics: `Textures appropriate to the system’s era and material reality: aged parchment, cast iron, printed paper currency, injection-moulded plastic, oxidised coin metal; realistic weight and scale :: 3` Illumination: `Neutral documentary lighting — diffused overhead, slight warm key, minimal shadow, clean background that does not editorialize :: 2` Render Stack: `Cinema 4D render, front-facing, 8k, historically-grounded material suite :: 1` Negative: `[Illegible, 2D, flat, political propaganda aesthetic, floating, cartoon, anachronistic objects] :: -1`
The best part is when your image is ready, you can edit each part.
[HT]
