Atlas

Compare Models

Claude Opus 5 vs GPT-6.2 Astra

GPT-6.2 Astra leads by 4.6 capability points (≈95% interval +2.0 to +7.2): a clear lead over Claude Opus 5 · P(higher) ≈ 100%

Wins 46 of 55 shared benchmarks (9 lost)

Capability vs. relative cost
Claude Opus 5GPT-6.2 AstraClosed weightsOpen weights

Benchmark wins9 – 46 · 55 shared

Comparing matching setups only (15 benchmarks excluded) · suites counted once
← Claude Opus 5GPT-6.2 Astra →
Data: Artificial Analysis, Epoch AI, Vals AI and 12 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-6.2 Astra minus Claude Opus 5.

Terminal-Bench-Science 0.1Science › Multidisciplinary28.678.1+49.5
EBR-bench v4 (card ban, single agent)Games45.792.4+46.7
Terminal-Bench Science 0.1 (Vals)Science › Multidisciplinary27.171.4+44.3
Mystery Game PuzzlesGames59.095.0+36.0
Chess Puzzles (task version 2.0)Games43.078.0+35.0
Humanity's Sixth SensePerception & media › Spatial & visual reasoning30.563.2+32.7
Blueprint-Bench 2Perception & media › Spatial & visual reasoning0.30.6+31.0
DeepsecBench — Revalidated V1 (233 issues)Software › Cybersecurity32.361.2+28.9
Furniture Assembly BenchmarkPerception & media › Spatial & visual reasoning60.889.2+28.3
ARC-AGI-3 (Semi-Private)Games30.258.0+27.8
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge44.872.2+27.4
FrontierMath Tier 4 (v2) - PrivateMath73.2100.0+26.8
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine13.939.4+25.6
AutomationBench 1.0.6 (Zapier)Tool & computer use › Customer service & operations26.952.4+25.4
EnterpriseBench: CoreCraftTool & computer use › Customer service & operations68.743.6−25.1
TaxCalcBench TY25Professional › Finance, accounting & tax20.044.0+24.0
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax23.845.8+22.0
DeepsecBench-AA (Artificial Analysis)Software › Cybersecurity25.946.8+20.9
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations56.677.2+20.7
Vals Science IndexScience › Multidisciplinary47.968.4+20.5
MirrorCode (ML, +Private, 2L)Software › Software engineering87.867.8−20.0
Tax Agent BenchProfessional › Finance, accounting & tax76.757.9−18.8
Artificial Analysis Intelligence Index v5.0.1Composite indexes44.262.1+18.0
EnterpriseOps-Gym-AATool & computer use › Customer service & operations47.529.9−17.6
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents6.315.2+17.5
MysteryMechanismMath37.453.2+15.8
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering49.064.1+15.2
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering31.245.4+14.2
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents10.337.0+13.4
Legal Research BenchProfessional › Legal55.342.8−12.5
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine22.034.4+12.4
CritPt v2 (agentic, avg@5)Science › Physical sciences71.483.7+12.3
FrontierMath ErdősMath0.010.3+10.3
GPQA DiamondScience › Multidisciplinary87.996.8+9.0
AA-AnalystAgent (pass^5)Professional › General knowledge work53.862.5+8.8
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work21.629.8+8.2
FrontierMath Tiers 1-3 (v2) - PrivateMath85.692.6+7.0
Epoch Capabilities Index (ECI)Composite indexes165173+6.5
Chartography (November 2026 rescoring)Perception & media › Documents & charts84.490.7+6.3
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge37.650.0+6.2
LMArena Agent Arena Steering BurdenTool & computer use › General agents2.414.0+5.8
MMMU ProKnowledge & reasoning › General knowledge84.089.5+5.5
ARC-AGI-2Knowledge & reasoning › Abstract reasoning90.495.8+5.4
TxBench Preclinical Pharmacology V1 Clean RerunsScience › Life sciences63.068.0+5.0
BigFinanceBench private holdout (final-answer accuracy)Professional › Finance, accounting & tax49.254.0+4.8
LLM Debate BenchmarkLanguage & writing › Language & wordplay17041673−4.4
SimpleBenchKnowledge & reasoning › General knowledge80.677.3−3.3
Vibe Code Bench v1.1Software › Frontend & design88.489.9+1.5
BioMysteryBench (Vals Terminus 2)Science › Life sciences79.380.7+1.5
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents10.112.4+1.2
OTIS Mock AIME 2024-2025Math98.9100.0+1.1
ARC-AGI-1 Semi-PrivateKnowledge & reasoning › Abstract reasoning97.598.5+1.0
AA-Briefcase v1.1 EloProfessional › General knowledge work16541644−0.6
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents8.77.7−0.5
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling1.01.2+0.1
Data: Artificial Analysis, Epoch AI, Vals AI and 12 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…