Meta Releases Muse Spark 1.3 with Major Agentic and Coding Gains

On September 3, Meta officially released Muse Spark 1.3, available the same day on Muse Code and the Meta Model API. Meta and executive Alexandr Wang positioned it as the strongest model in the Spark series to date, with a focus on real-world usability for agentic and coding tasks.

Confirmed

Not Yet Confirmed

Why It Matters

The core narrative of this upgrade is a shift from single-turn capability toward practical, long-horizon agentic workflows: fewer tool calls and lower token consumption mean real cost and latency reductions for agent tasks—another signal that model vendors' competition is pivoting to agent scenarios.

2026-09-03 ~ 2026-09-03 · 16 related posts

Primary sources

3 near-duplicate retellings: AIatMeta · ns123abc · alexandr_wang