Atlas

Compare Models

GPT-6 Astra vs Qwen4-Max

GPT-6 Astra leads by 3.4 capability points (≈95% interval +0.8 to +6.1): a clear lead over Qwen4-Max · P(higher) ≈ 99%

Wins 29 of 45 shared benchmarks (16 lost)

Capability vs. relative cost
GPT-6 AstraQwen4-MaxClosed weightsOpen weights

Benchmark wins29 – 16 · 45 shared

Comparing matching setups only (5 benchmarks excluded) · suites counted once
← GPT-6 AstraQwen4-Max →
Data: Artificial Analysis, Surge AI, Vals AI and 6 othersatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Qwen4-Max minus GPT-6 Astra.

Terminal-Bench-Science 0.1Science › Multidisciplinary54.823.3−31.4
EnterpriseBench: CoreCraftTool & computer use › Customer service & operations42.668.7+26.1
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & medicine41.719.4−22.2
Code Migration v1.1 (Vals Index v2.2 subset)Software › Software engineering78.057.3−20.8
DAYJOB: Law (November 2026 release)Professional › Legal36.617.5−19.1
Humanity's Last Exam DiamondKnowledge & reasoning › General knowledge63.844.9−18.9
CritPt v2 (agentic, avg@5)Science › Physical sciences83.466.0−17.4
sudo L7 (October 2026 rescoring)Software › Software engineering31.115.6−15.5
FrontierCode 2.0 (Artificial Analysis)Software › Software engineering41.426.5−14.9
LMArena Agent Arena Net Improvement (Steering Burden)Tool & computer use › General agents14.67.3−14.2
HLAB-Vals-1.1Professional › Legal7.121.3+14.2
Artificial Analysis Intelligence Index v5.0.1Composite indexes56.942.9−14.0
Chartography (November 2026 rescoring)Perception & media › Documents & charts88.675.3−13.3
DAYJOB: Finance (October 2026 rescoring)Professional › Finance, accounting & tax37.325.3−12.0
ComplexConstraints (November 2026 rescoring)Language & writing › Instruction following70.859.1−11.7
LiveBench 2.0 (2026-11-06 release)Composite indexes84.373.1−11.2
Terminal-Bench 4.0 (Artificial Analysis)Software › Software engineering59.648.5−11.1
Terminal-Bench 4.0Software › Software engineering59.648.5−11.1
GDPval-AA v2.1Professional › General knowledge work15721809+10.4
LLM Debate BenchmarkLanguage & writing › Language & wordplay16071679+10.3
LMArena Agent Arena Praise vs ComplaintTool & computer use › General agents32.313.0−9.6
AA-Omniscience v1.1 IndexKnowledge & reasoning › General knowledge44.627.7−8.4
HANDBOOK.md Agents v1.1Tool & computer use › Customer service & operations31.439.4+8.0
AA-Briefcase v1.1 EloProfessional › General knowledge work15441686+7.8
DAYJOB: AccountingProfessional › Finance, accounting & tax30.022.4−7.6
GDP.pdf (November 2026 rescoring)Professional › General knowledge work50.443.6−6.8
Tax Agent BenchProfessional › Finance, accounting & tax63.370.0+6.7
Agents' Last Exam v1.1 (ALE v1.1)Professional › General knowledge work40.246.7+6.6
Vals Finance IndexProfessional › Finance, accounting & tax55.662.0+6.4
Finance Agent v2.1Professional › Finance, accounting & tax60.466.7+6.3
GDP.pdf (Artificial Analysis, All-pass)Professional › General knowledge work32.226.2−6.0
AA-AnalystAgent (pass^5)Professional › General knowledge work51.346.3−5.0
AutomationBench-AA 1.0.6Tool & computer use › Customer service & operations68.572.6+4.1
DAYJOB: TaxProfessional › Finance, accounting & tax4.88.8+4.0
DAYJOB: Healthcare (October 2026 rescoring)Health & medicine28.424.4−4.0
LMArena Agent Arena Confirmed SuccessTool & computer use › General agents14.98.8−3.0
LMArena Agent Arena Steering BurdenTool & computer use › General agents13.98.0−3.0
FrontierCUA v1 pass@1Tool & computer use › Computer & browser use20.017.5−2.5
Vals Legal IndexProfessional › Legal68.470.7+2.3
DualEntry Accounting AI BenchmarkProfessional › Finance, accounting & tax81.283.2+2.0
Legal Research BenchProfessional › Legal39.441.3+1.9
Excel Modeling Benchmark v1.1Professional › Finance, accounting & tax72.770.8−1.9
LMArena Agent Arena Bash ReliabilityTool & computer use › General agents6.53.4−1.6
GDP.xlsxProfessional › General knowledge work22.923.7+0.8
LMArena Agent Arena Tool HallucinationTool & computer use › Tool calling1.21.20.0
Data: Artificial Analysis, Surge AI, Vals AI and 6 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…