Atlas

Compare Models

DeepSeek V4.1 Flash 1017 vs GPT-6.2 Astra

GPT-6.2 Astra leads by 8.2 capability points (≈95% interval +5.3 to +11.0): a clear lead over DeepSeek V4.1 Flash 1017 · P(higher) ≈ 100%

Wins 40 of 52 shared benchmarks (12 lost)

Capability vs. relative cost
DeepSeek V4.1 Flash 1017GPT-6.2 AstraClosed 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.

DeepSeek V4.1 Flash 1017vsGPT-6.2 Astracapability vs. relative costData Nov 17, 2026Village LLM Atlasatlas.village.dev

Benchmark wins12 – 40 · 52 shared

Comparing matching setups only · suites counted once
← DeepSeek V4.1 Flash 1017GPT-6.2 Astra →

DeepSeek V4.1 Flash 1017vsGPT-6.2 Astrawins by categoryData Nov 17, 2026Data: Artificial Analysis, Vals AI, Epoch AI and 4 othersVillage LLM Atlasatlas.village.dev

Margins

Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-6.2 Astra minus DeepSeek V4.1 Flash 1017.

ARC-AGI-3 (Semi-Private)General QA5.899.8+94.0
CyberGym-E2E-AA (Artificial Analysis)Cybersecurity76.30.0−76.3
AA-PersonalityBench Organization indexChat & Writing88.616.6−72.7
Terminal-Bench-Science 0.1Science12.978.1+65.2
FrontierMath Tier 4 (v2) - PrivateMath46.3100.0+53.7
Humanity's Last Exam DiamondGeneral QA20.872.2+51.4
AA-PersonalityBench Transparency indexChat & Writing86.736.1−51.1
AA-PersonalityBench Informativeness indexChat & Writing71.422.0−50.0
Mystery Game PuzzlesGames49.095.0+46.0
AA-PersonalityBench Warmth indexChat & Writing27.767.8+40.6
AA-PersonalityBench Slop indexChat & Writing39.11.0−38.5
SimpleQA Verified (Epoch anti-abstention)General QA42.578.5+36.0
Chess Puzzles (task version 2.0)Games46.078.0+32.0
CritPt v2 (agentic, avg@5)Science52.383.7+31.4
Terminal-Bench 4.0Code31.863.1+31.3
Terminal-Bench 4.0 (Artificial Analysis)Code33.864.1+30.3
Artificial Analysis Intelligence Index v5.0.1Indexes32.662.1+29.5
Artificial Analysis Engineering IndexIndexes37.666.6+29.0
AA-Omniscience v1.1 IndexIndexes1.550.0+24.2
AA-PersonalityBench Imagination indexChat & Writing37.260.4+23.5
Artificial Analysis Economics IndexIndexes38.561.1+22.6
DeepsecBench-AA (Artificial Analysis)Cybersecurity24.946.8+21.9
FrontierMath Tiers 1-3 (v2) - PrivateMath71.292.6+21.4
Artificial Analysis Legal IndexIndexes37.758.7+21.0
Artificial Analysis Finance & Accounting IndexIndexes42.260.9+18.7
ARC-AGI-2General QA78.195.8+17.8
Tax Agent BenchFinance & Accounting74.957.9−17.1
FrontierCode 2.0 (Artificial Analysis)Code28.545.4+16.9
Artificial Analysis Healthcare & Medical IndexIndexes41.357.0+15.7
Code Migration v1.1 (Vals Index v2.2 subset)Code68.583.6+15.1
HealthBench Professional (Length-Adjusted)Health & Medicine51.766.3+14.6
Artificial Analysis Strategy & Ops IndexIndexes49.263.3+14.1
AA-PersonalityBench Conviction indexChat & Writing94.380.9−13.5
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine26.739.4+12.8
GDP.pdf (Artificial Analysis, All-pass)Professional Work17.229.8+12.6
Epoch Capabilities Index (ECI)Indexes159174+11.8
Agents' Last Exam v1.1 ScoreAgents66.678.3+11.7
MMMU ProMultimodal79.789.5+9.8
AutomationBench-AA 1.0.6Agents68.577.2+8.8
AA-Briefcase v1.1 EloAgents14161570+8.4
AA-PersonalityBench Expressiveness indexChat & Writing52.460.0+7.7
Excel Modeling Benchmark v1.1Finance & Accounting69.276.0+6.9
Vals Index v2.2Indexes61.367.7+6.4
CWE-Bench-AA (Artificial Analysis)Cybersecurity69.265.0−4.2
ARC-AGI-1 Semi-PrivateGeneral QA94.598.5+4.0
Vals Finance IndexIndexes60.957.3−3.6
Legal Research BenchLegal45.242.8−2.4
Finance Agent v2.1Finance & Accounting62.960.7−2.2
Vals Legal IndexIndexes71.969.7−2.2
Vibe Code Bench v1.1Code87.789.9+2.2
HLAB-Vals-1.1Legal14.216.3+2.1
OTIS Mock AIME 2024-2025Math99.4100.0+0.6

DeepSeek V4.1 Flash 1017vsGPT-6.2 Astranormalised score marginsData Nov 17, 2026Data: Artificial Analysis, Vals AI, Epoch AI and 4 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…