Meta introduced Muse Spark 1.3 in early September 2026 as part of ongoing work on personal agent capabilities, according to Axios coverage published on the same day. The model is tracked on Artificial Analysis as a proprietary release from Meta, sitting within a comparison hub that breaks the family into tiered variants. That aggregator page lists a "max" variant with an Intelligence Index score of 53 and an "xhigh" variant scoring 52, giving prospective users a sense of how the tiers trade reasoning depth against cost and throughput. The framing in coverage around personal agents suggests Meta is positioning Muse Spark 1.3 for interactive, multi-step assistant experiences rather than purely batch generative tasks.

From a practical standpoint, Muse Spark 1.3 fits users who want a Meta-originated model with documented tiering and external benchmarking visibility, making it easier to weigh it against other proprietary options on intelligence, output speed, and per-task cost. The presence of distinct max and xhigh variants indicates flexibility for callers that need to balance capability and budget, while continued coverage tied to personal agent development points to forward momentum in agent-style use cases. Teams evaluating the model for production assistants or agent pipelines can lean on the Artificial Analysis comparison pages for ongoing intelligence and speed tracking as the ecosystem matures.