Claude Fable 5.2 vs GPT-6.2 Astra
Claude Fable 5.2 leads by 1.3 capability points (≈95% interval −1.7 to +4.2): statistically tied · P(higher) ≈ 80%
Wins 45 of 73 shared benchmarks (25 lost, 3 tied)
Capability vs. relative cost
Claude Fable 5.2GPT-6.2 AstraClosed weightsOpen weights
Benchmark wins45 – 25 · 73 shared, 3 tied
Comparing matching setups only (46 benchmarks excluded) · suites counted once
Area← Claude Fable 5.2GPT-6.2 Astra →
Data: Artificial Analysis, Surge AI, Epoch AI and 17 othersatlas.village.dev
Margins
Each benchmark scaled 0–100% (its declared range, else the field’s range); Δ is GPT-6.2 Astra minus Claude Fable 5.2.
| KradleBench | Games | 37.8 | 78.3 | +40.5 |
| AA-PersonalityBench Transparency index | Language & writing › Style & personality | 81.0 | 40.9 | −40.4 |
| AA-PersonalityBench Organization index | Language & writing › Style & personality | 54.3 | 21.8 | −32.9 |
| SpireClimb | Games | 15.3 | 41.3 | +26.0 |
| EnterpriseBench: CoreCraft | Tool & computer use › Customer service & operations | 68.2 | 43.6 | −24.6 |
| LMArena Agent Arena Net Improvement (Steering Burden) | Tool & computer use › General agents | 27.4 | 15.2 | −23.8 |
| SWE Atlas v1.1 - Overall | Software › Software engineering | 80.5 | 60.5 | −20.1 |
| TaxCalcBench TY25 | Professional › Finance, accounting & tax | 62.0 | 44.0 | −18.0 |
| SimpleBench | Knowledge & reasoning › General knowledge | 94.9 | 77.3 | −17.6 |
| LMArena Agent Arena Praise vs Complaint | Tool & computer use › General agents | 68.7 | 37.0 | −15.8 |
| AA-PersonalityBench Informativeness index | Language & writing › Style & personality | 46.7 | 31.3 | −15.5 |
| SimpleQA Verified (Epoch anti-abstention) | Knowledge & reasoning › General knowledge | 94.0 | 78.5 | −15.5 |
| AA-PersonalityBench Warmth index | Language & writing › Style & personality | 82.9 | 68.1 | −15.0 |
| DAYJOB: Law (November 2026 release) | Professional › Legal | 26.5 | 41.4 | +14.9 |
| AA-PersonalityBench Expressiveness index | Language & writing › Style & personality | 81.0 | 66.5 | −14.6 |
| sudo L7 (October 2026 rescoring) | Software › Software engineering | 52.2 | 38.3 | −13.9 |
| HLAB-Vals-1.1 | Professional › Legal | 5.0 | 16.3 | +11.3 |
| DAYJOB: Healthcare (October 2026 rescoring) | Health & medicine | 45.0 | 34.4 | −10.6 |
| Chartography (November 2026 rescoring) | Perception & media › Documents & charts | 80.8 | 90.7 | +9.9 |
| Tax Agent Bench | Professional › Finance, accounting & tax | 67.0 | 57.9 | −9.1 |
| Finance Agent v2.1 | Professional › Finance, accounting & tax | 69.5 | 60.7 | −8.7 |
| AutomationBench 1.0.6 (Zapier) | Tool & computer use › Customer service & operations | 44.3 | 52.4 | +8.1 |
| Humanity's Last Exam Diamond | Knowledge & reasoning › General knowledge | 80.2 | 72.2 | −8.0 |
| AA-Briefcase v1.1 Elo | Professional › General knowledge work | 1785 | 1644 | −7.7 |
| MultiChallenge | Language & writing › Conversation | 80.4 | 73.2 | −7.2 |
| RiemannBench (November 2026 rescoring) | Math | 83.2 | 76.0 | −7.2 |
| HANDBOOK.md Agents v1.1 | Tool & computer use › Customer service & operations | 52.3 | 45.8 | −6.5 |
| DAYJOB: Tax | Professional › Finance, accounting & tax | 15.6 | 9.2 | −6.4 |
| AA-3D v1.0 Elo | Software › Frontend & design | 1271 | 1318 | +6.3 |
| Vals Finance Index | Professional › Finance, accounting & tax | 63.6 | 57.3 | −6.3 |
| Professional Reasoning Benchmark - Legal | Professional › Legal | 49.0 | 43.1 | −5.9 |
| GDPval v2 (OpenAI internal) | Professional › General knowledge work | 44.8 | 50.4 | +5.6 |
| Agents' Last Exam v1.1 Score | Professional › General knowledge work | 72.8 | 78.3 | +5.5 |
| LMArena Agent Arena Steering Burden | Tool & computer use › General agents | 24.5 | 14.0 | −5.3 |
| ComplexConstraints (November 2026 rescoring) | Language & writing › Instruction following | 64.5 | 69.6 | +5.1 |
| LMArena Agent Arena Confirmed Success | Tool & computer use › General agents | 22.5 | 12.4 | −5.0 |
| Dual-Arm Sim (task completion) | Tool & computer use › Robotics | 7.5 | 2.5 | −5.0 |
| Mystery Game Puzzles | Games | 90.0 | 95.0 | +5.0 |
| Blueprint-Bench 2 | Perception & media › Spatial & visual reasoning | 0.6 | 0.6 | +4.8 |
| Medical Long Context Reasoning (MLCR-AA, October 2026 methodology) | Health & medicine | 43.9 | 39.4 | −4.4 |
| AA-PersonalityBench Imagination index | Language & writing › Style & personality | 67.6 | 63.3 | −4.4 |
| AA-PersonalityBench Conviction index | Language & writing › Style & personality | 75.3 | 71.3 | −4.0 |
| Vibe Code Bench v1.1 | Software › Frontend & design | 93.8 | 89.9 | −3.9 |
| Humanity's Sixth Sense | Perception & media › Spatial & visual reasoning | 59.5 | 63.2 | +3.7 |
| DAYJOB: Accounting | Professional › Finance, accounting & tax | 45.2 | 42.0 | −3.2 |
| GDP.xlsx | Professional › General knowledge work | 32.9 | 29.7 | −3.2 |
| AutomationBench-AA 1.0.6 | Tool & computer use › Customer service & operations | 74.2 | 77.2 | +3.1 |
| DualEntry Accounting AI Benchmark | Professional › Finance, accounting & tax | 81.2 | 78.2 | −3.0 |
| GPQA Diamond | Science › Multidisciplinary | 93.9 | 96.8 | +3.0 |
| BioMysteryBench (Vals Terminus 2) | Science › Life sciences | 83.7 | 80.7 | −3.0 |
| ARC-AGI-3 (Semi-Private) | Games | 97.3 | 100.0 | +2.7 |
| LMArena Agent Arena Bash Reliability | Tool & computer use › General agents | 13.1 | 7.7 | −2.7 |
| GDP.pdf (November 2026 rescoring) | Professional › General knowledge work | 51.8 | 49.2 | −2.6 |
| CritPt v2 (agentic, avg@5) | Science › Physical sciences | 86.3 | 83.7 | −2.6 |
| FrontierMath Tiers 1-3 (v2) - Private | Math | 95.1 | 92.6 | −2.5 |
| LLM Debate Benchmark | Language & writing › Language & wordplay | 1688 | 1673 | −2.1 |
| DAYJOB: Finance (October 2026 rescoring) | Professional › Finance, accounting & tax | 43.7 | 45.8 | +2.1 |
| ARC-AGI-2 | Knowledge & reasoning › Abstract reasoning | 93.8 | 95.8 | +2.1 |
| Ramp SWE-Bench v2 (Bash harness) | Software › Software engineering | 73.0 | 75.0 | +2.0 |
| EBR-bench v4 (card ban, single agent) | Games | 90.5 | 92.4 | +1.9 |
| Gray Swan IPI Attack Success (k=15) | Safety & alignment | 4.1 | 2.4 | +1.7 |
| FrontierCode 2.0 (Artificial Analysis) | Software › Software engineering | 47.6 | 49.0 | +1.4 |
| MysteryMechanism | Math | 51.8 | 53.2 | +1.4 |
| Excel Modeling Benchmark v1.1 | Professional › Finance, accounting & tax | 77.4 | 76.0 | −1.3 |
| Chess Puzzles (task version 2.0) | Games | 79.0 | 78.0 | −1.0 |
| Furniture Assembly Benchmark | Perception & media › Spatial & visual reasoning | 90.0 | 89.2 | −0.8 |
| AA-PersonalityBench Slop index | Language & writing › Style & personality | 6.7 | 7.4 | +0.7 |
| Epoch Capabilities Index (ECI) | Composite indexes | 173 | 173 | −0.5 |
| OTIS Mock AIME 2024-2025 | Math | 99.7 | 100.0 | +0.3 |
| LiveBench 2.0 (2026-11-06 release) | Composite indexes | 84.9 | 84.9 | 0.0 |
| LMArena Agent Arena Tool Hallucination | Tool & computer use › Tool calling | 1.2 | 1.2 | 0.0 |
| ARC-AGI-1 Semi-Private | Knowledge & reasoning › Abstract reasoning | 98.5 | 98.5 | 0.0 |
| FrontierMath Tier 4 (v2) - Private | Math | 100.0 | 100.0 | 0.0 |
Data: Artificial Analysis, Surge AI, Epoch AI and 17 othersatlas.village.dev