Atlas

Compare Models

GPT-6.2 Astra vs Qwen4-Max

GPT-6.2 Astra leads by 4.9 capability points (≈95% interval +2.2 to +7.6): a clear lead over Qwen4-Max · P(higher) ≈ 100%

Wins 44 of 54 shared benchmarks (10 lost)

Capability vs. relative cost
GPT-6.2 AstraQwen4-MaxClosed weightsOpen weights

Benchmark wins44 – 10 · 54 shared

Comparing matching setups only (3 benchmarks excluded) · suites counted once
← GPT-6.2 AstraQwen4-Max →
Data: Artificial Analysis, Surge AI, Vals AI and 7 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Qwen4-Max minus GPT-6.2 Astra.

KradleBenchGames78.311.0−67.3
Terminal-Bench-Science 0.1Science › Multidisciplinary78.623.3−55.2
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge72.244.9−27.3
Code Migration v1.1 (Vals Index v2.2 subset)Software › Software engineering83.657.3−26.3
EnterpriseBench: CoreCraftTool & computer use › Customer service & operations43.668.7+25.1
DAYJOB: Law (November 2026 release)Professional › Legal41.417.5−23.9
EnterpriseOps-Gym-AATool & computer use › Customer service & operations25.348.5+23.2
sudo L7 (October 2026 rescoring)Software › Software engineering38.315.6−22.7
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering49.026.5−22.5
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax45.825.3−20.5
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine39.419.4−20.0
DAYJOB: AccountingProfessional › Finance, accounting & tax42.022.4−19.6
Artificial Analysis Intelligence Index v5.0.1Composite indexes62.142.9−19.2
CritPt v2 (agentic, avg@5)Science › Physical sciences83.166.0−17.1
Artificial Analysis Engineering IndexScience › Engineering66.650.1−16.6
AA-AnalystAgent (pass^5)Professional › General knowledge work62.546.3−16.2
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering64.148.5−15.7
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents15.27.3−15.5
Chartography (November 2026 rescoring)Perception & media › Documents & charts90.775.3−15.4
FrontierCUA v1 pass@1Tool & computer use › Computer & browser use32.517.5−15.0
Terminal-Bench 4.0Software › Software engineering63.148.5−14.6
Tax Agent BenchProfessional › Finance, accounting & tax57.970.0+12.2
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents37.013.0−12.0
LiveBench 2.0 (2026-11-06 release)Composite indexes84.973.1−11.8
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge50.027.7−11.1
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following69.659.1−10.5
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine34.424.4−10.0
Artificial Analysis Economics IndexComposite indexes61.151.3−9.8
Artificial Analysis Legal IndexProfessional › Legal58.749.6−9.2
Artificial Analysis Healthcare & Medical IndexHealth & medicine57.048.2−8.8
Artificial Analysis Finance & Accounting IndexProfessional › Finance, accounting & tax60.954.1−6.8
Artificial Analysis Strategy & Ops IndexProfessional › General knowledge work63.356.9−6.4
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations45.839.4−6.4
GDP.xlsxProfessional › General knowledge work29.723.7−6.0
Finance Agent v2.1Professional › Finance, accounting & tax60.766.7+5.9
GDP.pdf (November 2026 rescoring)Professional › General knowledge work49.243.6−5.6
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax76.070.8−5.2
AA-Briefcase v1.1 EloProfessional › General knowledge work15941686+5.0
DualEntry Accounting AI BenchmarkProfessional › Finance, accounting & tax78.283.2+5.0
HLAB-Vals-1.1Professional › Legal16.321.3+5.0
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations77.272.6−4.7
Vals Finance IndexProfessional › Finance, accounting & tax57.362.0+4.7
Vals Index v2.2Composite indexes67.763.2−4.5
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work29.826.2−3.6
Vibe Code Bench v1.1Software › Frontend & design89.986.7−3.2
LMArena Agent Arena Steering BurdenTool & computer use › General agents14.08.0−3.0
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents7.73.4−2.2
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents12.48.8−1.8
Legal Research BenchProfessional › Legal42.841.3−1.4
Vals Legal IndexProfessional › Legal69.770.7+1.0
Agents' Last Exam v1.1 (ALE v1.1)Professional › General knowledge work47.546.7−0.8
LLM Debate BenchmarkLanguage & writing › Language & wordplay16731679+0.8
DAYJOB: TaxProfessional › Finance, accounting & tax9.28.8−0.4
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling1.21.20.0
Data: Artificial Analysis, Surge AI, Vals AI and 7 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…