GPT-5.2 Pro: benchmark results
OpenAI's higher-compute GPT-5.2 variant for difficult reasoning and general tasks. Provider: OpenAI. Released 2025-12-11. Access: API.
Aggregate Rating 1807 ± 11, rank #61 of 1656 rated models, from 152 benchmark results.
Strongest benchmark results
| Benchmark | Score | Metric | Percentile |
|---|---|---|---|
| ArmBench-LLM - MCQA | 98.67 | Accuracy (%) | 100 |
| ArmBench-LLM - NER | 78.35 | Accuracy (%) | 100 |
| ArmBench-LLM - Text Processing | 87.91 | Accuracy (%) | 100 |
| AutoBench | 4.48 | Judge Rating (1-5) | 100 |
| ConstraintBench | 65 | Feasibility rate (%): share of the 200 tasks whose returned | 100 |
| ConstraintBench (Joint Optimality) | 30.5 | Joint optimality rate (%): share of the 200 tasks whose solu | 100 |
| ConstraintBench (Joint Optimality) - Bin Packing | 40 | Joint optimality rate (%): share of the domain's 20 tasks wh | 100 |
| ConstraintBench (Joint Optimality) - Job-Shop Scheduling | 70 | Joint optimality rate (%): share of the domain's 20 tasks wh | 100 |
| ConstraintBench (Joint Optimality) - Order Fulfillment | 25 | Joint optimality rate (%): share of the domain's 20 tasks wh | 100 |
| ConstraintBench (Joint Optimality) - Production Mix | 90 | Joint optimality rate (%): share of the domain's 20 tasks wh | 100 |
| ConstraintBench (Joint Optimality) - Shift Scheduling | 55 | Joint optimality rate (%): share of the domain's 20 tasks wh | 100 |
| ConstraintBench - Bin Packing | 70 | Feasibility rate (%): share of the domain's 20 tasks whose r | 100 |
Interactive version: theaggregate.ai/model?slug=gpt-5-2-pro · How It Works · Data refreshed daily, snapshot 2026-10-20.