Tencent Hy4 Becomes Top Open-Weights Model on APEX-Agents

Key Info

Tencent Hy4 has become the top open-weights model on the APEX-Agents benchmark, scoring 77.6% Pass@1 and 85.1% mean score, ranking third among all 69 models evaluated.

Highlights

• Hy4 leads open-weights models with 77.6% Pass@1 (#3 of 69 overall)
• Mean score of 85.1% also ranks #3 among all 69 models
• Next open-weights competitor DeepSeek-V4.1-Flash-1017 reaches 69.8% Pass@1
• Kimi K3.1 Pro (65.5%), Qwen3.8-Max (63.3%), and MiMo V2.6 Flash round out the open-weights top 5