Meta Global AI Developer Hackathon. $1M in prizes.Apply Now
Muse Spark 1.1
Everything your build needs in a single, highly capable model. Now in public preview for US developers.Meet Muse Spark
End-to-end agentic workflows
Advanced coding capabilities
Native multimodal perception
Muse Spark benchmarks
Muse Spark is trained to deliver competitive agentic and coding performance, with multimodal perception built in.Benchmark
Muse Spark 1.1Meta
Muse SparkMeta
Gemini 3.1 ProGoogle
Opus 4.8Anthropic
GPT 5.5OpenAI
Agents
MCP AtlasScaled tool use
88.1
82.2
78.2
82.2
75.3
JobBenchProfessional tool use
54.7
17.0
15.9
48.4
38.3
Toolathlon-VerifiedPersonal tool use
75.6
49.4
61.1
76.2
73.5
OSWorld-VerifiedAgentic computer use
80.8
53.3
76.2
83.4
78.7
Humanity's Last ExamMultidisciplinary reasoning (w/ tools)
62.1
50.4
51.4
57.9
52.2
Finance Agent v2Agentic financial analysis
57.2
-
43.0
53.9
51.8
Coding
Terminal-Bench 2.1Agentic terminal coding
80.0
67.3
70.3
82.7
83.4
SWE-Bench ProDiverse software engineering
61.5
55.0
54.2
69.2
58.6
DeepSWE 1.1Long-horizon agentic coding
53.3
10.0
12.0
59.0
67.0
Multimodal
CharXiv ReasoningChart QA
88.4
88.9
81.6
89.9
84.8
BabyVisionVisual reasoning
76.3
39.9
51.5
81.2
83.6
For more details about evaluations, see our report.
Cookbooks and quickstarts
Meta Model API
Build with Muse Spark, now available on Meta Model API | AI Developers blogGet started with Muse Spark on Meta Model API: how to make your first call, the coding primitives the model is tuned for, and the agentic patterns that get the most out of it.Read more