Benchmarks/Muse Spark 1.2
Meta / Proprietary

Muse Spark 1.2

Meta’s recent frontier model, showing a large gain on agentic professional work.

INTELLIGENCE54index
CODINGNOT MEASURED
AGENTIC1631Elo
SPEEDNOT MEASURED
FIT / INTERPRETATION

Where the evidence points.

  • Agentic knowledge work
  • Tool use
  • Complex workflows
TECHNICAL CONTEXT

Score needs conditions.

Context
Not disclosed
Input / 1M
Not recorded
Output / 1M
Not recorded
Access
Proprietary
EVIDENCE TRAIL

Original measurements.

Muse Spark 1.2 evaluationOPEN SOURCE ↗