GPT-5.2 (High): benchmark results

GPT-5.2 evaluated at the high reasoning-effort setting. Provider: OpenAI. Released 2025-12-11. Access: API.

Unified ELO 1691 ± 1, rank #127 of 1761 rated models, from 156 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
CLEM AdventureGame99.17Game Clemscore (%)100
CLEM Clean Up100Game Clemscore (%)100
CLEM Codenames87.69Game Clemscore (%)100
CLEM Deal or No Deal99.12Game Clemscore (%)100
CLEM GuessWhat93.33Game Clemscore (%)100
GDPval (OpenAI Evals)49.88Win Rate (%)100
LLM2014 Logic 2025-1281.83Median Score100
LLM2014 Logic 2026-0180.71Median Score100
LiveCodeBench Pro Hard15.94Pass@1 (%)100
LiveCodeBench Pro Medium52.11Pass@1 (%)100
MCPMark57.48Pass@1 (%)100
Pencil Puzzle Bench - Yajilin20Direct-ask Success Rate (%)100

Interactive version: theaggregate.ai/model?slug=gpt-5-2-high · How It Works · Data refreshed daily, snapshot 2026-09-05.