GPT-5.6 Luna — benchmark results

Provider: OpenAI. Released 2026-07-09. Access: API.

Unified ELO 1917 ± 29, rank #54 of 1744 rated models, from 59 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA Long Context Reasoning74Accuracy (%)98.5
Artificial Analysis Intelligence Index51.24Intelligence Index97.4
Vals AI IOI72.92Accuracy (%)96.6
AA CritPt20.57Accuracy (%)96.3
AA GPQA Diamond91.11Accuracy (%)96.2
AA Omniscience - Software Engineering (SWE) - TypeScript77.78Accuracy (%)95.9
AA Humanity's Last Exam37.21Accuracy (%)95.8
AA SciCode52.55Accuracy (%)95.8
OTIS Mock AIME 2024-2598.33Accuracy (%)95.8
GDPval-AA1592Elo95.2
AA GDPval1591.81ELO95
Vals AI Terminal-Bench 2.179.03Accuracy (%)94.9

Interactive version: aibenchmarks.dev/model?slug=gpt-5-6-luna · How the rankings work · Data refreshed daily, snapshot 2026-07-10.