Ring-2.6-1T — benchmark results

InclusionAI's open 1T-parameter reasoning MoE (63B active), the reasoning counterpart to Ling-2.6-1T. Provider: InclusionAI. Released 2026-05-14. Access: Open.

Unified ELO 1697 ± 30, rank #273 of 1776 rated models, from 43 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
AA TAU-2 Bench92.4Accuracy (%)87.9
CritPt3.7Accuracy (self-reported)87.6
AA GPQA Diamond85.66Accuracy (%)85.1
AA CritPt3.71Accuracy (%)83.3
AA-LCR64.3Score (self-reported)82.3
AA Omniscience - Software Engineering (SWE) - Swift60Accuracy (%)80.9
AA SciCode42.36Accuracy (%)80.4
AA Long Context Reasoning64.33Accuracy (%)79.8
AA Omniscience - Software Engineering (SWE) - HTML48Accuracy (%)77.7
AA Humanity's Last Exam18.26Accuracy (%)77.6
Artificial Analysis Intelligence Index30.57Intelligence Index76.2
AA Omniscience - Science, Engineering & Mathematics34.9Accuracy (%)76

Interactive version: theaggregate.ai/model?slug=ring-2-6-1t · How the rankings work · Data refreshed daily, snapshot 2026-07-22.