Meta Muse Wonder Advances in the Frontier AI Race

Meta has launched Muse Wonder and Muse Wonder Full, its first frontier-class models from Meta Superintelligence Labs and its most powerful models to date. Both support a context window of up to one million tokens and combine cross-conversation memory with persistent task focus for AI agents. They accept text, images, video and audio, and were trained at Meta’s Prometheus data centre. Meta says Muse Wonder Full outperformed GPT-6 Astra and Claude Opus 5.5 on several internal benchmarks, including coding, legal tasks and memory. Independent evaluator Artificial Analysis gives Wonder Full a score of 56 and Muse Wonder 52, compared with 58 for Opus 5.5. Meta says the models also performed strongly in finance, healthcare and long-running software development tests. Muse Wonder is available through Meta Model API and Muse Code for $4 per million input tokens and $20 per million output tokens. Muse Wonder Full is initially limited to Muse Code subscribers; Meta has not announced when broader free access will begin. The new memory features may prompt data-privacy questions, and Meta says a full safety report is forthcoming. The launch is an AI-sector development, not a direct cryptocurrency-market catalyst, and the summaries identify no specific crypto asset affected.