Meta Unveils New Series of Advanced A.I. Models
Also covered by 18 sources
What happened
Meta has released two new AI models, Muse Wonder and Muse Wonder Full, through its Muse Code subscription and Meta Model API. These models are designed to handle complex personal and professional tasks, including legal, financial, and health-related work. The launch comes after 14 months of development by Meta Superintelligence Labs and follows Meta's previous model, Muse Spark. Muse Wonder and Wonder Full are now powering Meta's Muse personal assistant and are available to paid users.
Key details
- Muse Wonder Full achieved an Intelligence Index v4.3 score of 56, while Muse Wonder scored 52, both higher than Muse Spark 1.3's 48.
- The models offer a 1 million token context window and accept text, image, video, and audio inputs, with text-only output.
- Pricing is set at $4 per million input tokens and $20 per million output tokens for Muse Wonder; Wonder Full is exclusive to higher-tier Muse Code plans.
- Wonder Full ranked in the top three on 6 out of 10 industry evaluation benchmarks, including #1 on GDP.pdf and CritPt.
- A launch-day bug caused abnormal quota usage for subscribers, which was fixed and quotas were reset the same day.
Why it matters
Businesses and professionals using AI for legal, financial, and health tasks now have access to Meta's most advanced models, which can process larger and more complex workloads than before. Meta's Muse personal assistant and its API clients gain improved capabilities, including better memory and coding performance, at a cost competitive with other leading AI providers. Companies deciding between AI platforms face new options for advanced agentic tasks, with pricing and performance benchmarks that directly compete with OpenAI and Anthropic's latest offerings.