Muse Spark 1.3
Muse Spark 1.3 is a multimodal reasoning model from Meta for long-running agentic, multi-agent, and coding workflows. It improves long-horizon agent collaboration, instruction following, and coding efficiency relative to Muse Spark 1.2.
Shogo Agent Index
Estimated#28 of 9254.7range 41.6–67.7 · 17% of weight measured
Agentic coding
60
26% measured
Long-horizon autonomy
64
estimated
Knowledge-work agents
45
32% measured
Computer use
43
estimated
Security & context learning
53
estimated
Agentic benchmarks
| Benchmark | Score | Normalised | Index weight | Harness / source |
|---|---|---|---|---|
| SWE-bench Verified | est. | 8% | predicted 81 | |
| Terminal-Bench 2.0 | est. | 8% | predicted 82 | |
| DeepSWE | est. | 6% | predicted 58 | |
| FrontierSWE | 25.8% | 5% | proximus · $243.34/task | |
| CursorBench | 41.6% | 4% | source · $2.64/task | |
| GSO-Bench | est. | 4% | predicted 41 | |
| METR Time Horizons | est. | 10% | predicted 79 | |
| Vending-Bench 2 | est. | 6% | predicted 64 | |
| PostTrainBench | est. | 4% | predicted 28 | |
| APEX-Agents | 58.6% | 8% | source | |
| GDPval | est. | 7% | predicted 53 | |
| Remote Labor Index | est. | 5% | predicted 6 | |
| DeepResearch Bench | est. | 5% | predicted 49 | |
| OSWorld 2.0 | est. | 6% | predicted 15 | |
| OSWorld | est. | 4% | predicted 84 | |
| Cybench | est. | 5% | predicted 82 | |
| CL-bench | est. | 5% | predicted 24 |
Other benchmarks
| Benchmark | Category | Score | Source |
|---|---|---|---|
| Chess Puzzles | Reasoning | 37.0% | source |
| CritPt | Reasoning | 26.0% | source |
| DTBench | Knowledge | 96.5% | source |
| FrontierMath Tiers 1-3 | Math | 74.4% | source |
| FrontierMath Tier 4 | Math | 46.3% | source |
| GeoBench | Multimodal | 86.0% | source |
| LMCA | Knowledge | 53.9% | source |
| MirrorCode | Coding | 54.4% | source |
| OTIS Mock AIME | Math | 99.2% | source |
| ProofBench | Math | 58.0% | source |
| SciCode | Coding | 59.7% | source |
| SimpleBench | Reasoning | 81.8% | source |
| WebDev Arena | Coding | 1652 | source |
| WeirdML v3 | Coding | 6.7% | codex_cli |