Meta 首个前沿模型 Muse Wonder：内部测试对标 GPT-6 Astra
Meta 的 Muse Wonder 内部测试与 GPT-6 Astra、Fable 5.1 持平，MemoryBench 上超过 Fable、Opus 和 Astra，定价每百万输入 $4、输出 $20，尚无独立评测。
Do you realize what Zuckerberg just confessed about Meta's first frontier model? Mark Zuckerberg: "We're gonna write this check for like $100 billion, $200 billion. I'm like, Alex, are you sure this is going to work?" Alexandr Wang: "…certainly not clear at all that it was going to work." The model is Muse Wonder. What most people are missing: Meta's internal tests place it level with GPT-6 Astra and Fable 5.1. On MemoryBench it comes out ahead of Fable, Opus and Astra. Pricing is $4 per million input tokens and $20 per million output, the same tier as Opus 5.5. It was trained at Prometheus, Meta's new Ohio data center. There are no independent benchmarks yet. One bet worth $200 billion, and the person signing for it wasn't sure it would pay off. Now we see whether it did. Interested in the latest with ai/tech and business? Follow @GeniusGTX and never fall behind. I track ChatGPT, Claude, and how every new agent is changing how we work and think, then hand you the pure signals.