List your company

General Perception and Reasoning on MME-RealWorld Lite

54.3Overall AccuracyDeepeyes
38.49242.59646.750.804May 20, 2025Jul 18, 2025Sep 16, 2025Nov 14, 2025Jan 13, 2026Mar 13, 2026May 12, 2026

Evaluation results

MethodLinks
DeepeyesParameter Size=-2026.0354.3—————————50.556.6
DeepEyesReasoning Type=Tool-based Visual Reasoning2026.0554.3—————————50.556.6
V-ReflectionParameter Size=8B2026.0353.9—————————4558.5
DeepEyesParam Size=7B2025.0553.29052.78943.333.476694435——
GPT-4oParameter Size=-2026.0352—————————48.354.4
GPT-4oReasoning Type=Textual Reasoning2026.0552—————————48.354.4
UniVLRReasoning Type=Visual Latent Reasoning2026.0550.7—————————44.754.5
Qwen2.5-VL-7BParameter Size=7B, SFT=vanilla2026.0350.1—————————42.554.1
Pixel-ReasonerParam Size=7B2025.0549.789.6528638.930.971724632.5——
PixelReasonerReasoning Type=Tool-based Visual Reasoning2026.0549.7—————————44.553.1
InternVL3Parameter Size=8B2026.0348.6—————————44.851
CoVTReasoning Type=Visual Latent Reasoning2026.0548.2—————————42.951.6
MonetReasoning Type=Visual Latent Reasoning2026.0546.9—————————40.351.2
Qwen2.5-VL-7BReasoning Type=Textual Reasoning2026.0546.2—————————43.148.2
Qwen2.5-VL-7BParameter Size=7B2026.0345.8—————————39.749.6
Qwen2.5-VLParam Size=32B2025.0545.687.240.78329.540.7746027.329.5——
SkiLaReasoning Type=Visual Latent Reasoning2026.0545.6—————————36.351.6
Qwen2.5-VL-7B + vanilla SFTReasoning Type=Textual Reasoning, SFT status=vanilla SFT2026.0545.5—————————39.949.1
LLaVA-OneVisionParam Size=7B2025.0543.780405631.739.465333832——
Qwen2.5-VLParam Size=7B2025.0542.387.632.78327.330726228.723——
LVRReasoning Type=Visual Latent Reasoning2026.0539.1—————————37.540.1