• •Meta launched Muse Spark 1.3, which reports 20% fewer tool calls and 25% fewer tokens compared to version 1.2, according to MarkTechPost.
• •TechCrunch reports Meta is offering financial incentives or discounts to users who opt into sharing model usage data for product development.
• •While MarkTechPost emphasizes the technical efficiency gains of the new model, it does not detail the data-sharing policy mentioned by TechCrunch.
• •It remains unclear how Meta validates the quality of the shared data or how the model performs in complex, long-horizon coding environments outside of standard benchmarks.

Meta released Muse Spark 1.3 this week, an agentic coding model designed to reduce computational overhead by lowering tool call volume. While MarkTechPost focuses on the model’s internal efficiency improvements over version 1.2, TechCrunch notes a shift in business strategy by offering users discounts in exchange for access to their usage patterns. This creates a clear divide between the model's technical performance and the privacy trade-offs required to lower operating costs. Whether these incentives will be enough to attract professional developers concerned about data security remains to be seen.