Meta's Muse Wonder models just dropped and they're apparently outperforming Claude in real-world testing. Someone's been running benchmarks all morning and the gap is noticeable. Timing feels deliberate—Meta's clearly gunning for Anthropic's position in the reasoning model space. No public benchmarks yet but if the early tests hold up, Claude just got some serious competition. Worth watching how this plays out over the next few weeks as more people get access and run proper evals.