Muse Spark 1.1 — benchmark results

Meta's Muse Spark 1.1 model, an updated release of the closed Muse Spark system. Provider: Meta. Released 2026-07-09. Access: Private.

Unified ELO 1886 ± 18, rank #72 of 1776 rated models, from 67 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
LLM Stats (MCP Atlas)88.1Score (%)100
LLM Stats (Toolathlon)75.6Score (%)100
SEAL - MCP Atlas88.1Score100
SEAL - MultiNRC65.59Score100
SEAL - Professional Reasoning Benchmark - Finance55.01Score100
SEAL - Professional Reasoning Benchmark - Legal57.05Score100
SEAL - SWE-Bench Pro (Private Dataset)51.5Score100
SEAL - SWE-Bench Pro (Public Dataset)61.5Score100
SEAL - VisualToolBench (VTB)44.77Score100
Vals AI Harvey Legal Agent Bench20Accuracy (%)100
Vals AI MedScribe88.89Accuracy (%)100
Vals AI TaxEval v279.72Accuracy (%)100

Interactive version: theaggregate.ai/model?slug=muse-spark-1-1 · How the rankings work · Data refreshed daily, snapshot 2026-07-22.