Unknown · 3× · vu le 07 août 2026
30
Momentum
Reve (also known as Reve Image, currently version 2.0/2.1) is a text-to-image model by Reve AI, Inc., a startup based in Palo Alto, California. It employs a "layout-first" approach: prompts are first translated into a structured, code-like image layout representation with position, size, and description of each element before being natively rendered in 4K (4096x4096 pixels, 16 megapixels). The model achieved 2nd place on the Arena Leaderboard for text-to-image behind OpenAI's GPT Image 2, with access via web app (app.reve.com) and a credit-based API (api.reve.com).
Historique du momentum
09.05.07.08.
Fonctionnalités
| Fine-Tuning | No public fine-tuning; Free/Lite/Pro content is used for model training by default (opt-out available) |
| Generation Time | Avg. 86.8 seconds (Image Editing Arena, Reve 2.0) |
| Max Resolution | Native 4K (4096x4096 pixels, 16 megapixels) |
| Platform | Web app (app.reve.com) and API (api.reve.com) |
| Price | Web app: Free / Lite $7.99 per month / Pro $19.99 per month; API: from $10 for 7,500 credits, approx. $0.024–0.04 per image (Create/Edit/Remix) |
| Release Date | Reve 2.0 released on June 3, 2026 |
