Meta Muse Wonder Full Matches Sonnet 5.5 on Artificial Analysis Benchmark
Meta’s new Muse Wonder Full model scores 56 on the Intelligence Index, matching Sonnet 5.5 and surpassing Fable 5.1 at maximum reasoning effort.
Key points
- Meta released Muse Wonder and Muse Wonder Full models with improved benchmark scores.
- Muse Wonder Full scored 56 on the Artificial Analysis Intelligence Index v4.3 at max reasoning effort.
- The new models show a material gain over the previous Muse Spark 1.3 model released six weeks prior.
Meta has released two new artificial intelligence models, Muse Wonder and Muse Wonder Full, achieving higher scores on the Artificial Analysis Intelligence Index v4.3. According to a report by Techmeme, the larger Muse Wonder Full model scored 56 at maximum reasoning effort, while the standard Muse Wonder model scored 52.
These results place Muse Wonder Full on par with Anthropic’s Sonnet 5.5 (max) and slightly behind Opus 5.5 (max). The new models outperform Fable 5.1 (max) on the same index. This performance represents a material gain over Meta’s earlier Muse Spark 1.3 (max) model, which scored 48 on the benchmark.
View the original post on X
Benchmark Performance and Iteration Speed
The scoring data comes from Artificial Analysis, which tracks model performance on its Intelligence Index. According to the outlet, Muse Wonder Full achieved a score of 56 at max reasoning effort, matching the performance of Sonnet 5.5 (max). The standard Muse Wonder model also improved significantly, reaching a score of 52 under the same conditions.
These scores reflect the performance metrics reported by Artificial Analysis. The release gap between the previous Muse Spark 1.3 (max) and the new Muse Wonder models was only six weeks. During this period, the benchmark scores increased from 48 to 52 and 56 respectively, with the source describing this as a material gain.
Model Availability and Context
Muse Wonder is available for users on Meta’s first-party platforms, according to the source material. The specific availability details for the Muse Wonder Full variant were not explicitly detailed in the provided text, though both models were announced as released. The comparison benchmarks include other major industry models such as Fable 5.1 (max), Sonnet 5.5 (max), and Opus 5.5 (max).
The performance metrics highlight the competitive landscape for large language models. By matching Sonnet 5.5 (max) and approaching Opus 5.5 (max), Meta positions its latest offerings against top-tier competitors. The outperformance of Fable 5.1 (max) further underscores the improvements in the latest Muse series.
What should you do?
Security and technology professionals should monitor the deployment of these new models in production environments. Evaluate whether the increased reasoning capabilities require updated safety guardrails or new monitoring protocols. Review access controls for any systems integrating Muse Wonder or Muse Wonder Full to ensure appropriate authorization levels.
Background: Large language models
Large language models are a specific type of neural network trained on text. Generative AI is the broader umbrella term for any system that creates new content, including images and code. Not all generative AI uses large language models, but most modern text generators do.
What to do and how to stay safe: Meta
- Review integration plans for Muse Wonder and Muse Wonder Full to assess security implications of increased reasoning capabilities.
- Monitor official channels from Meta and Artificial Analysis for detailed performance breakdowns and potential vulnerabilities.
- Evaluate current benchmarking strategies to include comparisons with new models like Muse Wonder Full against existing baselines.
- Ensure access controls are updated for any applications deploying these new Meta models to prevent unauthorized usage.
General security guidance from the Threat Times newsroom. It is not confirmed advice from the organisations named in this story.
Frequently asked questions
How does Muse Wonder Full compare to Sonnet 5.5?
According to Artificial Analysis, Muse Wonder Full matches Sonnet 5.5 (max) with a score of 56 on the Intelligence Index v4.3 at max reasoning effort.
What was the score of the previous Muse Spark 1.3 model?
Muse Spark 1.3 (max) previously scored 48 on the Artificial Analysis Intelligence Index, which is lower than the new Muse Wonder models.
How quickly were these new models released?
The new Muse Wonder models were released only six weeks after the Muse Spark 1.3 (max) model, showing a rapid iteration cycle.
