A new or updated AI model from a lab: what it can do, what it costs, how it compares.
Meta's Muse Wonder models score close to the best from OpenAI and Anthropic in tests. Running them costs nearly three times what Meta's older model did.
