Muse Wonder Full (max) outperforms Fable 5.1 (max) on Artificial Analysis' Intelligence Index, matches Sonnet 5.5 (max), and is just shy of Opus 5.5 (max)

Meta's improved LLMs narrow the gap with leading models, potentially offering developers more competitive performance choices.

Meta's new language models, Muse Wonder and Muse Wonder Full, have been evaluated by Artificial Analysis. Muse Wonder Full achieved a score of 56 on the Intelligence Index v4.3 at maximum reasoning effort, a significant improvement from the previous Muse Spark 1.3 (max) which scored 48. Muse Wonder scored 52. This performance places Muse Wonder Full just below Anthropic's Opus 5.5 (max) and on par with Anthropic's Sonnet 5.5 (max). It also outperforms Fable 5.1 (max). The substantial gains were achieved within a six-week release gap, indicating rapid development from Meta in the LLM space. Artificial Analysis noted that faster token generation is becoming increasingly critical for complex, multi-step agentic workflows in AI.

Meta's rapid iteration cycle, with Muse Wonder Full showing significant gains over its predecessor in just six weeks, highlights an accelerating pace of development in the frontier LLM market. This fierce competition benefits users by pushing the boundaries of AI capabilities and potentially driving down costs as models become more efficient. The focus on token generation speed also signals a shift towards AI agents capable of more complex, autonomous task completion.

The performance of Muse Wonder Full on the Artificial Analysis Intelligence Index suggests that Meta is closing the gap with industry leaders like Anthropic. Developers and businesses relying on LLM APIs should monitor these advancements, as Meta's competitive offerings could influence pricing and feature sets across the sector. The increasing importance of agentic workflows implies a future where AI performs more sophisticated, multi-step tasks, requiring robust and fast model performance.