Muse Spark 1.3: benchmark results
Meta's Muse Spark 1.3 (September 2026), a multimodal reasoning model for long-running agentic and multi-agent coding workflows, accepting text, images, video, audio and files with a 1,048,576-token context. Provider: Meta. Released 2026-09-02. Access: API.
Aggregate Rating 1870 ± 7, rank #24 of 1640 rated models, from 401 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| Design Arena (Website) | 1360 | Elo | 100 |
| Harvey LAB-AA - All Pass | 26.67 | All Criteria Pass Rate (%) | 100 |
| Harvey LAB-AA - Antitrust & Competition | 95.17 | Criterion Pass Rate (%) | 100 |
| Harvey LAB-AA - Arbitration & International Dispute Resolution | 96.65 | Criterion Pass Rate (%) | 100 |
| Harvey LAB-AA - Banking & Finance | 97.64 | Criterion Pass Rate (%) | 100 |
| Harvey LAB-AA - Bankruptcy & Restructuring | 95.62 | Criterion Pass Rate (%) | 100 |
| Harvey LAB-AA - Capital Markets | 97.65 | Criterion Pass Rate (%) | 100 |
| Harvey LAB-AA - Corporate Governance | 97.37 | Criterion Pass Rate (%) | 100 |
| Harvey LAB-AA - Data Privacy & Cybersecurity | 98.12 | Criterion Pass Rate (%) | 100 |
| Harvey LAB-AA - Emerging Companies & Venture Capital | 96.7 | Criterion Pass Rate (%) | 100 |
| Harvey LAB-AA - Employment & Labor | 93.63 | Criterion Pass Rate (%) | 100 |
| Harvey LAB-AA - Environmental & ESG | 95.22 | Criterion Pass Rate (%) | 100 |
Interactive version: theaggregate.ai/model?slug=muse-spark-1-3 · How It Works · Data refreshed daily, snapshot 2026-10-13.