Tencent's Hy4 benchmarks close in on Claude Opus 5.5 and GPT 6.1 Sol, topping several leaderboards
heypearlai · x · 2026-11-05
Tencent has released Hy4, and third-party commentary calls it "decent": it doesn't beat the big three (Claude Opus 5.5, GPT 6.1 Sol, Kimi K3) overall, but the gap is getting uncomfortably close.
From the benchmark charts:
- Hy4 takes the top spot on OSWorld 2.1 (78.9) and leads several internal benchmarks like E-Bench 2.0 and Hy-BrowseComp 2.0
- It closely trails the frontier on coding/agent benchmarks like ProgramBench, FrontierCode 2.0 and Terminal Bench 4.0, clearly ahead of GLM 5.3 and DeepSeek-V4.1-Flash
- Weak spots remain: Terminal Bench Science 0.1 (19.8), PostTrainBench and Agents' Last Exam still lag the leaders
The author's takeaway: the gap between the absolute frontier and open models keeps shrinking, and Tencent is cooking.
More from Models
- BioSimBench: frontier LLMs ace biology benchmarks but fail at predicting real cell experiments — alex_verem · 2026-11-05
- Kimi K3.1 Pro gains less from test-time compute than Opus 5.0, data shows — teortaxesTex · 2026-11-05
- Grok and Claude Now Refuse to Root Your Own Devices, Citing Brick Risk — doooyle · 2026-11-05
- Nace AI offers 120% cashback to switch teams to Drex Max, claims 11x cost edge over Claude Code — testingcatalog · 2026-11-05
- Stanford benchmark: 18 AI models tested as teammates, best scores just 82.2/100 — james_y_zou · 2026-11-05
- Kimi K3.1 Pro hits 0.78x human compute multiplier, lagging frontier by ~4.6 months — ChowdhuryNeil · 2026-11-05