Math benchmarks: GPT-6.1 Sol leads as open models shine

Open-source model Inkling 2 Small topped MathArena's BrokenArXiv benchmark, while GPT-6.1 Sol (max) led math/proof benchmarks like ArXivMath and LeanLean with 90.43% accuracy, with low-cost models also performing well.

2026-11-12 ~ 2026-11-12 · 2 related posts

• Open model Inkling 2 Small takes top open spot on MathArena's BrokenArXiv — teortaxesTex · 2026-11-12
• Math Benchmark Sheet Shows 90.43% Top Accuracy at Just $1.21 per Run — teortaxesTex · 2026-11-12