Peanut - Text to Image Model (Open Weights coming soon)

A mysterious new player has entered the text-to-image arena—literally. Peanut, developed by an anonymous team, has debuted at #8 in the Artificial Analysis Text to Image Arena, a benchmark that ranks models by user preference and generation quality. What makes Peanut particularly noteworthy is its upcoming open-weight release, which the team confirmed is imminent. Once the weights become public, Peanut will overtake existing open-weight champions like Z-Image Turbo, Qwen-Image, and FLUX.2 [dev], establishing itself as the top open-weight text-to-image model available.

For developers and AI enthusiasts, this means access to state-of-the-art image generation without relying on proprietary APIs or restrictive licenses. Open weights allow fine-tuning, local deployment, and integration into custom pipelines—a major advantage for research, startup applications, and privacy-sensitive use cases. Peanut's exact architecture and training details remain under wraps, but its arena performance suggests strong alignment and visual fidelity. The anonymous nature adds an element of intrigue, but the priority is clearly on delivering a competitive open-source alternative. The community eagerly awaits the weight release to verify claims and explore capabilities.