Meta has launched Muse Spark 1.1, an updated AI model that outperforms the GLM-5.2 in coding tasks while reducing operational costs. According to The Decoder, Muse Spark 1.1 achieved a score of 71.3 in coding performance, surpassing GLM-5.2 at a lower cost of $0.26 per task.
The Decoder also reported that Muse Spark 1.1’s rating on the Artificial Analysis Intelligence Index rose by eight points over three months, reaching 51 points. Importantly, the model’s hallucination rate—a measure of generating inaccurate outputs—dropped significantly from 73 percent to 38 percent, reflecting improved reliability.
For Japanese markets, where AI-driven automation in FX and equities trading is gaining momentum, advancements like Muse Spark 1.1 could contribute to more cost-effective and accurate algorithmic strategies.