Share your thoughts, 1 month free Claude Pro on usSee more

Visual Question Answering on RealWorldQA (test)

79Accuracy

GPT5 mini

Updated 15d ago

Evaluation Results

Method	Links
GPT5 mini 2026.01		79	-
Qwen3-VL 32B 2026.01		78.4	-
Qwen3-VL 30B-A3B 2026.01		77.4	-
Gemini-2.5 Flash 2026.01		76	-
MMFineReason-8B 2026.01		75.6	-
GPT-4o 2024.10		75.4	-
MMFineReason-4B 2026.01		74.9	-
Qwen3-VL 8B 2026.01		73.5	-
MMR1 8B 2026.01		71	-
SSL4RL-7B (Hard-Contrastive) 2025.10		70.58	-
HoneyBee 8B 2026.01		70.5	-
Qwen2 VL 2024.10		69.7	-
OMR 7B 2026.01		69.4	-
SSL4RL-7B (Mask) 2025.10		68.88	-
Qwen2.5VL 2025.12		68.5	-
LLaVA-OV-7B 2026.01		68.4	-
LLaVA-OV-7B w/ CLI 2026.01		68.4	-
MMFineReason-2B 2026.01		68.2	-
LLaVA-OV-7B w/ SLI 2026.01		68.2	-
DiffusionVL 2025.12		68	-
InternVL-2-26B 2026.01		67.2	-
GPT-4o-mini 2024.10		67.1	-
LLaVA-OV 2025.12		66.3	-
Qwen2.5-VL-7B 2025.10		65.88	-
Qwen2.5VL 2025.12		65.4	-
InternVL-2-8B 2026.01		64.4	-
Cambrian-1 2025.12		64.2	-
MiniCPM-Llama3-V 2.5 2024.10		63.5	-
LLaDA-V 2025.12		63.2	-
Baichuan-omni 2024.10		62.6	-
DiffusionVL 2025.12		61.6	-
LLaVA-OV-7B w/ DeepStack 2026.01		61.4	-
VITA 2024.10		59	-
IXC-2.5-7B 2026.01		57.5	-
LLaVA-OV-0.5B w/ CLI 2026.01		56.7	-
Fine-Tuning (Clean) 2026.05		56.6	0
LLaVA-OV-0.5B 2026.01		56	-
LLaVA-OV-0.5B w/ SLI 2026.01		55.1	-
Zero-Shot (Clean) 2026.05		54.9	5.9
Zero-Shot (Clean) 2026.05		53.6	3.9
LLaVA-OV-0.5B w/ DeepStack 2026.01		52.9	-
MMGUARD-BPH 2026.05		52.9	3.7
MMGUARD-CRS 2026.05		51.7	4.9
Zero-Shot (Clean) 2026.05		49.7	6.9
VILA-13B 2026.01		41.9	-
Zero-Shot (Clean) 2026.05		36.6	19.6
Zero-Shot (Clean) 2026.05		27.5	39.8