List your company

Vision-Language Navigation on EmbodiedBench Navigation latest (test)

86.7Success RateGPT4.1+VP
13.932.851.770.6Dec 2, 2025

Evaluation results

MethodLinks
GPT4.1+VPBackbone=GPT4.12025.1286.7
VP+SRGPOBackbone=Qwen2.5-VL-3B-Ins2025.1272.3
Claude-3.5-SonnetAccess=Close-Source2025.1266.7
InternVL3-78BAccess=Open-Source2025.1266.7
GPT4.1Access=Close-Source2025.1265
Gemini-2.0-flashAccess=Close-Source2025.1263.3
Ovis2-34BAccess=Open-Source2025.1263.3
VP+GiGPO (w. VPR)Backbone=Qwen2.5-VL-3B-Ins2025.1257.2
GPT-4oAccess=Close-Source2025.1255
gemma-3-27b-itAccess=Open-Source2025.1253.3
Claude-3.7-SonnetAccess=Close-Source2025.1250
Qwen-VL-MaxAccess=Close-Source2025.1250
Llama-3.2-11B-Vision-InsAccess=Open-Source2025.1248.3
Qwen2.5-VL-72B-InsAccess=Open-Source2025.1246.7
VP+GRPOBackbone=Qwen2.5-VL-3B-Ins2025.1240.8
InternVL3-8BAccess=Open-Source2025.1238.3
VP+SFTBackbone=Qwen2.5-VL-3B-Ins2025.1236.7
GPT-4o-miniAccess=Close-Source2025.1231.7
VP+GiGPO (vanilla)Backbone=Qwen2.5-VL-3B-Ins2025.1229.4
Gemini-1.5-ProAccess=Close-Source2025.1223.3
Llama-3.2-90B-Vision-InsAccess=Open-Source2025.1223.3
Qwen2.5-VL-7B-InsAccess=Open-Source2025.1220
Qwen2.5-VL-3B-InsAccess=Open-Source2025.1216.7