Atlas

Compare Models

Claude Haiku 5.5 vs GPT-6.2 Astra

GPT-6.2 Astra leads by 7.8 capability points (≈95% interval +5.1 to +10.4): a clear lead over Claude Haiku 5.5 · P(higher) ≈ 100%

Wins 42 of 56 shared benchmarks (14 lost)

Capability vs. relative cost
Claude Haiku 5.5GPT-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.

Claude Haiku 5.5vsGPT-6.2 Astracapability vs. relative costData Nov 17, 2026Village LLM Atlasatlas.village.dev

Benchmark wins14 – 42 · 56 shared

Comparing matching setups only (2 benchmarks excluded) · suites counted once
← Claude Haiku 5.5GPT-6.2 Astra →

Claude Haiku 5.5vsGPT-6.2 Astrawins by categoryData Nov 17, 2026Data: Artificial Analysis, Vals AI, Epoch AI and 3 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 Claude Haiku 5.5.

ARC-AGI-3 (Semi-Private)General QA2.899.8+97.0
Mystery Game PuzzlesGames30.095.0+65.0
Terminal-Bench Science 0.1 (Vals)Science10.071.4+61.4
Terminal-Bench-Science 0.1Science20.078.1+58.1
SimpleQA Verified (Epoch anti-abstention)General QA23.878.5+54.7
AA-PersonalityBench Informativeness indexChat & Writing75.322.0−53.8
FrontierMath Tier 4 (v2) - PrivateMath46.3100.0+53.7
EBR-bench v4 (card ban, single agent)Games43.892.4+48.6
Humanity's Last Exam DiamondGeneral QA29.672.2+42.6
Furniture Assembly BenchmarkMultimodal47.589.2+41.7
AA-PersonalityBench Transparency indexChat & Writing77.236.1−41.5
AutomationBench-AA 1.0.6Agents36.077.2+41.3
AutomationBench 1.0.6 (Zapier)Agents12.052.4+40.3
ARC-AGI-2General QA61.395.8+34.6
Code Migration v1.1 (Vals Index v2.2 subset)Code50.283.6+33.4
Terminal-Bench 4.0 (Artificial Analysis)Code32.864.1+31.3
Vals Science IndexIndexes38.568.4+29.9
Terminal-Bench 4.0Code36.463.1+26.8
Artificial Analysis Intelligence Index v5.0.1Indexes36.562.1+25.7
AA-PersonalityBench Organization indexChat & Writing41.016.6−24.6
Artificial Analysis Strategy & Ops IndexIndexes39.163.3+24.2
Artificial Analysis Engineering IndexIndexes42.566.6+24.1
Chess Puzzles (task version 2.0)Games55.078.0+23.0
Artificial Analysis Legal IndexIndexes38.758.7+20.1
Artificial Analysis Healthcare & Medical IndexIndexes37.557.0+19.5
AA-Omniscience v1.1 IndexIndexes12.350.0+18.9
Artificial Analysis Economics IndexIndexes42.761.1+18.4
MysteryMechanismScience35.153.2+18.0
FrontierMath Tiers 1-3 (v2) - PrivateMath75.192.6+17.5
Artificial Analysis Finance & Accounting IndexIndexes43.660.9+17.3
AA-PersonalityBench Expressiveness indexChat & Writing77.260.0−17.3
CritPt v2 (agentic, avg@5)Science66.383.1+16.9
AA-PersonalityBench Conviction indexChat & Writing67.680.9+13.4
HLAB-Vals-1.1Legal4.216.3+12.1
MirrorCode (ML, +Private, 2L)Code78.967.8−11.1
Epoch Capabilities Index (ECI)Indexes160174+11.1
Excel Modeling Benchmark v1.1Finance & Accounting65.576.0+10.6
FrontierCode 2.0 (Artificial Analysis)Code34.945.4+10.4
MMMU ProMultimodal79.189.5+10.4
BioMysteryBench (Vals Terminus 2)Science70.480.7+10.4
AA-PersonalityBench Slop indexChat & Writing10.51.0−9.6
ARC-AGI-1 Semi-PrivateGeneral QA89.098.5+9.5
GDP.pdf (Artificial Analysis, All-pass)Professional Work20.829.8+9.0
Vals Index v2.2Indexes58.767.7+8.9
AA-AnalystAgent (pass^5)Professional Work53.862.5+8.8
Finance Agent v2.1Finance & Accounting65.460.7−4.6
Medical Long Context Reasoning (MLCR-AA, October 2026 methodology)Health & Medicine35.039.4+4.4
Tax Agent BenchFinance & Accounting62.257.9−4.3
AA-Briefcase v1.1 EloAgents15721644+3.9
AA-PersonalityBench Imagination indexChat & Writing63.860.4−3.5
AA-PersonalityBench Warmth indexChat & Writing69.567.8−1.7
OTIS Mock AIME 2024-2025Math98.9100.0+1.1
Legal Research BenchLegal43.842.8−1.0
Vals Legal IndexIndexes70.369.7−0.6
Vibe Code Bench v1.1Code90.489.9−0.5
Vals Finance IndexIndexes57.457.3−0.1

Claude Haiku 5.5vsGPT-6.2 Astranormalised score marginsData Nov 17, 2026Data: Artificial Analysis, Vals AI, Epoch AI and 3 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…