{"id":23706,"date":"2026-09-04T04:28:09","date_gmt":"2026-09-04T04:28:09","guid":{"rendered":"https:\/\/scannn.com\/introducing-muse-spark-1-3-meta-ai-research\/"},"modified":"2026-09-04T04:28:09","modified_gmt":"2026-09-04T04:28:09","slug":"introducing-muse-spark-1-3-meta-ai-research","status":"publish","type":"post","link":"https:\/\/scannn.com\/lv\/introducing-muse-spark-1-3-meta-ai-research\/","title":{"rendered":"Introducing Muse Spark 1.3 | Meta AI Research"},"content":{"rendered":"\n
\n

We\u2019re excited to release Muse\u00a0Spark\u00a01.3, which delivers improved performance across agentic and coding tasks. Drawing on what we learned from months of broad adoption of Muse\u00a0Code and Meta\u00a0Model\u00a0API, we\u2019ve also made this model easier to use in real-world settings. Smarter and more practically useful, Muse\u00a0Spark\u00a01.3 advances our work toward personal superintelligence.<\/p>\n

Muse\u00a0Spark\u00a01.3 is rolling out today in Muse\u00a0Code and Meta\u00a0Model\u00a0API. Previously available reasoning modes are available today with max reasoning coming shortly after we finish additional safety testing.<\/p>\n

\n
\n
<\/p>\n
<\/picture><\/div>\n<\/div>\n<\/div>\n<\/figure>\n

For more details about our evaluations, see our report<\/a>.<\/p>\n

Agentic Workflows<\/h2>\n

Muse\u00a0Spark\u00a01.3 is designed to better sustain longer-horizon work by collaborating with users and juggling multiple workflows in a single, long thread. When given an open-ended objective, it uses tools to generate its own context across messy and conflicting sources, proactively corrects gaps in its plan, and keeps track of what it has learned to produce a final deliverable. We trained the model across a diverse set of harnesses to generalize to various agentic environments.<\/p>\n

\n

Trained to more actively collaborate with the user, Muse\u00a0Spark\u00a01.3 asks clarifying questions when prompts are ambiguous, invokes help from the user when stuck, and confirms before taking consequential actions. When working on long tasks, it adapts to user preferences, either providing frequent updates or working silently in the background.<\/p>\n

Muse\u00a0Spark\u00a01.3 follows complex, long-form instructions more reliably than earlier Muse Spark models. Across multi-step tasks, it\u2019s better at preserving detailed requirements without dropping constraints or drifting from the requested workflow.<\/p>\n

\n
\n

Note: this AI agent prototype was created by Muse Spark and is not a real product<\/span>