GPT-5.2 ChatGPT: benchmark results

Provider: OpenAI. Released 2025-12-11. Access: API.

Unified ELO 1686 ± 1, rank #74 of 1392 rated models, from 8 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
HumainE Leaderboard32.12HumainE Score81.1
AI Chess Leaderboard (Continuation)842Elo75
VerdictBench63.44Accuracy (self-reported)75
Chess Bench LLM671Lichess Rating72.6
AI Chess Leaderboard (Reasoning)835Elo68.4
SpeechMap Compliance69.6% Requests Completed61.2
LLM Chess (Saplin)142.4ELO60.2
Bullshit Benchmark14.5BS Detection Rate (%)22.9

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