Meta’s Muse Wonder and Wonder Full are two new 5.5-tier models from MSL. Built on the Llama 3 architecture, they push 70B and 160B parameter scales with enhanced reasoning and multimodal capabilities. These models compete directly with Claude 3.5 and GPT-4o, offering open-weight alternatives for AI developers.
Opening Kapyn…