On September 2, local time, Meta released Muse Spark 1.3, stating that the new model improves performance in terms of agents and coding tasks, and enhances the ability to follow long-cycle tasks, multi-tasking, and complex instructions. The model has been launched with the Muse Code and Meta Model API, and the highest inference mode will be launched after completing security tests. According to Meta, compared with Muse Spark 1.2, Muse Spark 1.3 reduces the number of tool calls by about 20%, reduces token usage by about 25% in coding tasks, and enhances adversarial input and prompt injection protection capabilities. Meta is also planning to launch a larger model and open source weight for Muse Spark.

On September 2, local time, Meta released Muse Spark 1.3, stating that the new model improves performance in terms of agents and coding tasks, and enhances the ability to follow long-cycle tasks, multi-tasking, and complex instructions. The model has been launched with the Muse Code and Meta Model API, and the highest inference mode will be launched after completing security tests. According to Meta, compared with Muse Spark 1.2, Muse Spark 1.3 reduces the number of tool calls by about 20%, reduces token usage by about 25% in coding tasks, and enhances adversarial input and prompt injection protection capabilities. Meta is also planning to launch a larger model and open source weight for Muse Spark.