GPT-5 Nano: benchmark results

OpenAI's smallest, cheapest GPT-5 tier with a 400K context, built for high-volume classification, extraction and autocomplete (August 2025). Provider: OpenAI. Released 2025-08-07. Access: API.

Unified ELO 1567 ± 1, rank #376 of 1392 rated models, from 847 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
HELM Safety98.6Mean score (self-reported)100
PCB-Bench - QA Placement SBERT67.86SBERT similarity (%)100
PCB-Bench - QA Routing SBERT63.47SBERT similarity (%)100
StereoTales67Emissions (self-reported)100
OpenVLM MathVista - GPS93.8Accuracy (%)99.7
OpenVLM MMMU - Accounting86.7Accuracy (%)99.6
OpenVLM MMMU - Math76.7Accuracy (%)99.5
OpenVLM MathVista - SCI80.3Accuracy (%)99.3
OpenVLM MathVista - ALG89.7Accuracy (%)99
OpenVLM MathVista - GEO90.4Accuracy (%)99
OpenVLM MMMU - Mechanical Engineering63.3Accuracy (%)98.9
OpenVLM MMMU - Chemistry76.7Accuracy (%)98.8

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