Atlas

Compare Models

GPT-6.2 Sol vs Hy4

GPT-6.2 Sol leads by 5.1 capability points (≈95% interval +2.3 to +7.8): a clear lead over Hy4 · P(higher) ≈ 100%

Wins 10 of 11 shared benchmarks (1 lost)

Capability vs. relative cost
GPT-6.2 SolHy4Closed weightsOpen weights

Hy4 has no setting with a relative cost yet.

Benchmark wins10 – 1 · 11 shared

Comparing matching setups only (182 benchmarks excluded) · suites counted once
← GPT-6.2 SolHy4 →
Data: Surge AI, Artificial Analysis, OpenAI and 2 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Hy4 minus GPT-6.2 Sol.

Chartography (November 2026 rescoring)Perception & media › Documents & charts87.749.2−38.5
EnterpriseBench: CoreCraftTool & computer use › Customer service & operations36.468.2+31.8
RiemannBench (November 2026 rescoring)Math80.051.2−28.8
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge56.532.7−23.8
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax32.89.2−23.6
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations34.523.7−10.8
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine17.27.2−10.0
GDP.pdf (November 2026 rescoring)Professional › General knowledge work48.843.6−5.2
GDP.xlsxProfessional › General knowledge work22.317.7−4.6
Agents' Last Exam v1.1 ScoreProfessional › General knowledge work73.070.3−2.7
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following69.566.8−2.7
Data: Surge AI, Artificial Analysis, OpenAI and 2 othersatlas.village.dev
Win counts use each benchmark’s declared score direction. Represented suite slices and the derived Capability score are excluded from headline totals. Missing or incompatible setups are excluded, never counted as losses.
Loading Atlas data…