VulcanBench VulcanBench

Analysis

Deeper looks at how models and harnesses behave on VulcanBench tasks: where the time goes, how agents work, and what changes their cost and speed.

Analysis02
Muse Wonder Full against Claude Opus 5.5, GPT-6.1 Sol and GPT-6 Astra: combined score, time and cost
2026-10-15, updated 2026-10-17 · Morgan Linton · VulcanBench Frontier v4, Low to Extra-high effort · combined score, time and cost
→
Analysis01
Analysis of traces comparing Muse Wonder Full and GPT-6.1 Sol at Low effort
2026-10-14 · Morgan Linton · VulcanBench Frontier v4, Low effort · trace analysis
→
Read the benchmarks → Subscribe by RSS