Hy3-preview (Reasoning) — benchmark results

Tencent Hunyuan Hy3 preview (295B A21B open MoE) evaluated with reasoning enabled. Provider: Tencent. Released 2026-04-24. Access: Open.

Unified ELO 1769 ± 23, rank #168 of 1776 rated models, from 42 benchmark results.

Strongest benchmark results

BenchmarkScoreMetricPercentile
LLMEval-Logic Base75.3Accuracy (%)92.3
UGI - Natural Intelligence50.11NatInt Score91
UGI - Writing51.12Writing Score89.4
AA TAU-2 Bench92.69Accuracy (%)88.2
AA GPQA Diamond86.67Accuracy (%)87.3
AA Humanity's Last Exam25.53Accuracy (%)85.2
Wolfram LLM Benchmarking Project56.7Correct Functionality (%)85.2
AA Omniscience - Software Engineering (SWE) - Swift64Accuracy (%)84.9
AA CritPt4.57Accuracy (%)84.5
AA Omniscience - Science, Engineering & Mathematics37.3Accuracy (%)81.2
Artificial Analysis Intelligence Index33.58Intelligence Index80.6
AA Terminal-Bench Hard34.09Accuracy (%)78.8

Interactive version: theaggregate.ai/model?slug=hy3-preview-reasoning · How the rankings work · Data refreshed daily, snapshot 2026-07-22.