VECTOR WIREAI INTELLIGENCE
⌘K
UTC
Refresh Models —Deals —Regulatory —Sources —

Muse Spark vs Qwen3.8 Max Preview

24 SHARED BENCHMARKS

Across 24 shared benchmarks, Muse Spark scores higher on 6 and Qwen3.8 Max Preview on 18. The widest gap is TauBench V3 - Banking, where Qwen3.8 Max Preview scores 51.3 against 19.6.

METAVSALIBABA24 SHARED6–18 HEAD-TO-HEAD
BenchmarkMuse SparkMARGINQwen3.8 Max Preview
AA Agentic Index28.758.4
AA Intelligence44.358.1
AA-LCR7774.3
AA-Omniscience7.23.4
arena_elo14881481
arena_vision13061313
Artificial Analysis Coding Index58.671.8
coding_arena_elo15401669
critpt11.320
cybergym43.578.5
ERQA64.777.8
gdpval32.361
GPQA Diamond89.592.7
HLE58.443.6
IFBench75.982.8
MMMU-Pro80.582.3
OmniScience Accuracy49.631.9
OmniScience Non-Hallucination26.858.3
scicode51.552.9
screenspot_pro_no_tools84.184.5
simpleqa_verified66.345.8
SWE-bench Pro5567.7
TauBench V3 - Banking19.651.3
Terminal-Bench 2.162.286.6

Best tracked score per model per benchmark (default configuration; source-attributed). ↓ marks lower-is-better metrics. Open either model for its full surface, provenance and pricing.