Atlas

Compare Models

GPT-6.2 Astra vs GPT-6.2 Sol

GPT-6.2 Astra leads by 2.2 capability points (≈95% interval −0.8 to +5.2): statistically tied · P(higher) ≈ 93%

Wins 54 of 63 shared benchmarks (7 lost, 2 tied)

Capability vs. relative cost
GPT-6.2 AstraGPT-6.2 SolClosed weightsOpen weights

Each line runs through one model’s priced reasoning settings, cheapest to most expensive; grey points are other models’ priced settings in the same range. Results whose reasoning effort the source didn’t report are left out wherever a model has named effort levels. Capability and relative cost are the published estimates (relative cost is cost per task as a multiple of GPT-5 (high)); hover a setting, or use the arrow keys, to set it beside the other model’s same setting.

GPT-6.2 AstravsGPT-6.2 Solcapability vs. relative costData Nov 17, 2026Village LLM Atlasatlas.village.dev

Benchmark wins54 – 7 · 63 shared, 2 tied

Comparing matching setups only (2 benchmarks excluded) · suites counted once
← GPT-6.2 AstraGPT-6.2 Sol →

GPT-6.2 AstravsGPT-6.2 Solwins by categoryData Nov 17, 2026Data: Artificial Analysis, Vals AI, OpenAI and 2 othersVillage LLM Atlasatlas.village.dev

Margins

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

PostTrainBench Lite - release-reported performanceCode74.844.9−29.9
MysteryMechanismScience53.236.9−16.2
AA-PersonalityBench Organization indexChat & Writing16.630.8+14.3
Terminal-Bench Science 0.1 (Vals)Science71.458.6−12.9
AA-PersonalityBench Transparency indexChat & Writing36.147.4+11.4
AutomationBench 1.0.6 (Zapier)Agents52.441.3−11.1
DeepsecBench-AA (Artificial Analysis)Cybersecurity56.445.5−11.0
Vals Science IndexIndexes68.457.8−10.7
Terminal-Bench-Science 0.1Science78.167.6−10.5
SecureSWE (OpenAI internal)Cybersecurity65.755.6−10.1
ARC-AGI-2General QA95.885.8−10.0
AA-AnalystAgent (pass^5)Professional Work62.553.8−8.7
Humanity's Last Exam DiamondGeneral QA72.264.0−8.2
AA-3D v1.0 EloCode13181259−8.0
AA-PersonalityBench Warmth indexChat & Writing67.860.5−7.4
OpenAI research debugging - internal release scoreCode81.874.7−7.2
FrontierCode 2.0 (Artificial Analysis)Code49.042.8−6.2
AutomationBench-AA 1.0.6Agents77.271.3−6.0
Legal Research BenchLegal42.837.0−5.8
Artificial Analysis Engineering IndexIndexes66.661.2−5.5
Agents' Last Exam v1.1 ScoreAgents78.373.0−5.3
CritPt v2 (agentic, avg@5)Science83.778.6−5.1
Tax Agent BenchFinance & Accounting57.962.9+5.0
CWE-Bench-AA (Artificial Analysis)Cybersecurity69.264.2−5.0
Artificial Analysis Intelligence Index v5.0.1Indexes62.157.3−4.9
AA-Omniscience v1.1 IndexIndexes50.040.4−4.8
AA-PersonalityBench Conviction indexChat & Writing80.985.7+4.8
AA-PersonalityBench Expressiveness indexChat & Writing60.055.7−4.3
AA-PersonalityBench Imagination indexChat & Writing60.456.2−4.3
HLAB-Vals-1.1Legal16.312.1−4.2
PatchBench (OpenAI internal)Cybersecurity79.775.6−4.1
KernelGen1P - OpenAI release percentageCode76.974.7−4.1
Terminal-Bench 4.0 (Artificial Analysis)Code64.160.1−4.0
AA-PersonalityBench Informativeness indexChat & Writing22.025.9+4.0
Artificial Analysis Strategy & Ops IndexIndexes63.359.4−3.9
Artificial Analysis Economics IndexIndexes61.157.4−3.7
Artificial Analysis Legal IndexIndexes58.755.0−3.7
Artificial Analysis Healthcare & Medical IndexIndexes57.053.3−3.7
Artificial Analysis Finance & Accounting IndexIndexes60.957.2−3.7
GDPval v2 (OpenAI internal)Professional Work50.446.9−3.5
Vals Legal IndexIndexes69.766.7−3.0
ExploitBench - Internal Port (June-August 2026)Cybersecurity53.550.5−3.0
BioMysteryBench (Vals Terminus 2)Science80.777.8−3.0
Artificial Analysis Cyber Index v1Indexes40.637.7−2.9
Terminal-Bench 4.0Code63.160.6−2.5
MentalHealthBenchSafety53.350.9−2.4
LifeVal (GPT-6.2 launch chart)Agents86.183.8−2.3
Vibe Code Bench v1.1Code89.987.6−2.3
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine39.437.2−2.2
Finance Agent v2.1Finance & Accounting60.758.9−1.8
Code Migration v1.1 (Vals Index v2.2 subset)Code83.685.3+1.8
Excel Modeling Benchmark v1.1Finance & Accounting76.074.3−1.7
AA-Briefcase v1.1 EloAgents16441612−1.7
Vals Index v2.2Indexes67.766.1−1.6
ARC-AGI-3 (Semi-Private)General QA99.898.2−1.6
ARC-AGI-1 Semi-PrivateGeneral QA98.597.5−1.0
GPQA DiamondGeneral QA97.096.0−1.0
EnterpriseOps-Gym-AAAgents29.928.9−1.0
Vals Finance IndexIndexes57.356.7−0.6
HealthBench ProfessionalHealth & Medicine66.065.5−0.5
GDP.pdf (Artificial Analysis, All-pass)Professional Work29.830.0+0.2
AA-PersonalityBench Slop indexChat & Writing1.01.00.0
CyberGym-E2E-AA (Artificial Analysis)Cybersecurity0.00.00.0

GPT-6.2 AstravsGPT-6.2 Solnormalised score marginsData Nov 17, 2026Data: Artificial Analysis, Vals AI, OpenAI and 2 othersVillage LLM Atlasatlas.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…