List your company

Proactive Assistance on Pro2Bench HoloAssist

87G-Mean F1Llama 4-Maverick
-3.4820.0143.566.99Jun 3, 2026

Evaluation results

MethodLinks
Llama 4-MaverickFT=true, PWR=true, Architecture=split duplex+planner2026.068765
Qwen3.6-VL-27BFT=true, PWR=true, Architecture=split duplex+planner2026.067945
Llama 4-MaverickFT=true, Architecture=monolithic2026.067146
Gemini 3.1 ProArchitecture=monolithic, Mode=zero-shot2026.065341
Llama 4-MaverickArchitecture=monolithic2026.065047
Claude Opus 4.6Architecture=monolithic, Mode=zero-shot2026.064926
Qwen3.6-VL-27BFT=true, Architecture=monolithic2026.064931
Qwen3-VL-235BArchitecture=monolithic2026.064635
Qwen3.6-VL-27BArchitecture=monolithic2026.062241
GPT-5.2Architecture=monolithic, Mode=zero-shot2026.060—