Peanut debuts in image arena

The ranking is real signal, but the story is still incomplete until the weights and license ship. If Peanut is as strong in the wild as it looks in blind evals, it could reset expectations for open image models.

• –#8 overall is strong enough to matter immediately, especially in a benchmark crowded with closed models
• –The shared examples suggest it can handle both photorealistic scenes and stylized poster art, which are common failure modes for image generators
• –The biggest unknowns are model size, architecture, and licensing, which will decide whether this is a research curiosity or a practical open-weight release
• –If the weights do arrive cleanly, Peanut puts pressure on the current open-weight leaders to match both quality and usability
• –For teams tracking local or self-hosted image generation, this is the kind of model that can shift the shortlist fast