Atlas

Compare Models

GPT-6.2 Astra vs Grok 4.7

GPT-6.2 Astra leads by 8.4 capability points (≈95% interval +5.7 to +11.1): a clear lead over Grok 4.7 · P(higher) ≈ 100%

Wins 46 of 60 shared benchmarks (14 lost)

Capability vs. relative cost
GPT-6.2 AstraGrok 4.7Closed 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 AstravsGrok 4.7capability vs. relative costData Nov 17, 2026Village LLM Atlasatlas.village.dev

Benchmark wins46 – 14 · 60 shared

Comparing matching setups only (7 benchmarks excluded) · suites counted once
← GPT-6.2 AstraGrok 4.7 →

GPT-6.2 AstravsGrok 4.7wins by categoryData Nov 17, 2026Data: Artificial Analysis, Vals AI, Epoch AI and 1 otherVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is Grok 4.7 minus GPT-6.2 Astra.

ARC-AGI-3 (Semi-Private)General QA99.810.0−89.7
FrontierMath Tier 4 (v2) - PrivateMath100.017.1−82.9
CyberGym-E2E-AA (Artificial Analysis)Cybersecurity0.074.0+74.0
Furniture Assembly BenchmarkMultimodal89.220.8−68.3
Terminal-Bench-Science 0.1Science78.611.4−67.1
Mystery Game PuzzlesGames95.029.0−66.0
Terminal-Bench Science 0.1 (Vals)Science71.410.0−61.4
Humanity's Last Exam DiamondGeneral QA72.223.6−48.6
FrontierMath Tiers 1-3 (v2) - PrivateMath92.653.0−39.6
Terminal-Bench 4.0 (Artificial Analysis)Code64.125.8−38.4
Chess Puzzles (task version 2.0)Games78.041.0−37.0
ARC-AGI-2General QA95.861.4−34.4
Terminal-Bench 4.0Code63.128.8−34.3
Vals Science IndexIndexes68.434.8−33.6
Code Migration v1.1 (Vals Index v2.2 subset)Code83.651.8−31.8
CritPt v2 (agentic, avg@5)Science83.755.7−28.0
MysteryMechanismScience53.225.2−27.9
Artificial Analysis Intelligence Index v5.0.1Indexes62.135.5−26.7
DeepsecBench-AA (Artificial Analysis)Cybersecurity52.726.9−25.8
AA-PersonalityBench Imagination indexChat & Writing60.435.3−25.4
Artificial Analysis Engineering IndexIndexes66.641.8−24.8
AA-PersonalityBench Warmth indexChat & Writing67.844.8−23.3
SimpleQA Verified (Epoch anti-abstention)General QA78.556.0−22.5
AA-PersonalityBench Transparency indexChat & Writing36.158.1+22.3
FrontierCode 2.0 (Artificial Analysis)Code49.028.7−20.3
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine39.421.1−18.3
Artificial Analysis Cyber Index v1Indexes40.656.4+15.8
AA-PersonalityBench Expressiveness indexChat & Writing60.075.3+15.4
Artificial Analysis Economics IndexIndexes61.145.7−15.4
Epoch Capabilities Index (ECI)Indexes174156−14.1
MirrorCode (ML, +Private, 2L)Code67.854.4−13.3
AA-PersonalityBench Organization indexChat & Writing16.629.6+13.0
AutomationBench-AA 1.0.6Agents77.265.6−11.7
Artificial Analysis Legal IndexIndexes58.747.2−11.6
AA-PersonalityBench Informativeness indexChat & Writing22.033.4+11.5
BioMysteryBench (Vals Terminus 2)Science80.769.3−11.5
MMMU ProMultimodal89.578.2−11.3
Tax Agent BenchFinance & Accounting57.969.0+11.1
Artificial Analysis Healthcare & Medical IndexIndexes57.046.6−10.4
FrontierMath ErdősMath10.30.0−10.3
AA-Omniscience v1.1 IndexIndexes50.031.2−9.4
Artificial Analysis Finance & Accounting IndexIndexes60.951.7−9.2
Vals Index v2.2Indexes67.758.6−9.0
Artificial Analysis Strategy & Ops IndexIndexes63.354.8−8.5
ARC-AGI-1 Semi-PrivateGeneral QA98.590.2−8.3
AA-PersonalityBench Slop indexChat & Writing1.08.6+7.7
GDP.pdf (Artificial Analysis, All-pass)Professional Work29.823.2−6.6
AA-AnalystAgent (pass^5)Professional Work61.355.0−6.2
Excel Modeling Benchmark v1.1Finance & Accounting76.071.5−4.5
Legal Research BenchLegal42.847.1+4.3
GPQA DiamondGeneral QA96.892.7−4.1
Vibe Code Bench v1.1Code89.986.2−3.7
Vals Legal IndexIndexes69.772.7+3.0
AA-Briefcase v1.1 EloAgents15941636+2.3
HLAB-Vals-1.1Legal16.318.3+2.1
OTIS Mock AIME 2024-2025Math100.098.1−1.9
AA-PersonalityBench Conviction indexChat & Writing80.982.9+1.9
Vals Finance IndexIndexes57.358.8+1.5
CWE-Bench-AA (Artificial Analysis)Cybersecurity69.268.3−0.8
Finance Agent v2.1Finance & Accounting60.760.6−0.2

GPT-6.2 AstravsGrok 4.7normalised score marginsData Nov 17, 2026Data: Artificial Analysis, Vals AI, Epoch AI and 1 otherVillage 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…