Meta's Muse Spark 1.1 Scores 51 on the Artificial Analysis Intelligence Index (twitter.com)

Meta has announced the release of Muse Spark 1.1, which has achieved a score of 51 on the Artificial Analysis Intelligence Index, representing an 8-point improvement over its predecessor, Muse Spark 1.0. This score positions Muse Spark 1.1 competitively alongside leading models such as GLM-5.2 and GPT-5.4, just behind Grok 4.5 and Claude Fable 5. The enhancements are particularly notable in areas such as scientific reasoning, coding capabilities, and overall agentic knowledge work, although it still trails slightly in the latter category. The significance of Muse Spark 1.1 lies in its cost and efficiency benefits, making it one of the more affordable options in the market. With only 94 million output tokens required to run the Intelligence Index, it operates at approximately $0.26 per task, significantly cheaper than its rivals. Additionally, the model boasts a massive context window of 1 million tokens, improved from 262,000, and a median output speed of 114 tokens per second. The decrease in hallucination rates further underscores its reliability, with a notable reduction from 73% to 38%. Overall, Muse Spark 1.1 not only showcases advancements in AI capabilities but also represents a step towards more economical AI solutions for developers.