Meta Platforms

Artificial Intelligence · Enterprise Software

Launch

Meta launches Muse Wonder, its first frontier AI model to rival GPT-6, Claude

Pricing it on par with OpenAI and Anthropic's top models signals Meta now sees its AI as a premium product, not just a free perk.

• Meta launched Muse Wonder and Muse Wonder Full on October 12, 2026, calling them its first true frontier-class AI models built for personal and professional AI agents.
• The models come from Meta Superintelligence Labs, the unit CEO Mark Zuckerberg built after hiring former Scale AI CEO Alexandr Wang following Meta's $14 billion investment in Scale AI.
• Muse Wonder is available via the Meta Model API and Muse Code at $4 per million input tokens and $20 per million output tokens; Muse Wonder Full is reserved for Muse Code subscribers only.
• The models introduce 'Active Memory' to retain user context across conversations and 'Active Focus' to keep agents working toward a goal through errors and setbacks.
• On Artificial Analysis benchmarks, Meta says Muse Wonder Full outperforms OpenAI's GPT-6 Astra and Anthropic's Claude Fable 5.1, ranking just below Claude Opus 5.5.
• Both models were trained at Meta's gigawatt-scale Prometheus data center, following April's lighter Muse Spark release that marked Meta's pivot away from its underperforming open-source Llama line.
• Muse Wonder will soon power Meta's Muse personal assistant, rolling out first to paid subscribers before reaching its broader base of more than 3.6 billion daily active users.
• The release is the clearest test yet of whether Meta's hundreds of billions in AI spending, which has drawn investor skepticism, can actually close the gap with OpenAI and Anthropic.