SOTA Logical Reasoning on LogicVista and PapersWithCode

81.4Accuracy

Gemini3-pro

Updated 18d ago

Evaluation Results

Method	Links
Gemini3-pro 2026.05		81.4
Gemini-2.5-Pro 2026.07		73.8
Doubao-Seed-1.6 2026.05		72.5
Qwen3-VL-235B-Thinking 2026.05		72.3
GPT-5-20250807 2026.05		70
GPT5 2026.07		70
GLM-4.5V 2026.05		62.4
OST 2026.05		61.5
Qwen3-VL-8B-Instruct 2026.05		61.3
DEITA 2026.05		61.2
Qwen3-VL-8B-Instruct + LLM-as-a-Judge 2026.05		60.9
Qwen3-VL-8B-Instruct + Random 2026.05		60.6
OST 2026.05		60.4
Qwen3-VL-4B-Instruct 2026.05		60
Qwen3-VL-8B-Instruct + Full SFT 2026.05		60
Qwen3-VL-4B-Instruct + LLM-as-a-Judge 2026.05		59.7
DEITA 2026.05		59.6
Qwen3-VL-4B-Instruct + Random 2026.05		59.5
Qwen3-VL-30B A3B-Instruct 2025.12		58.2
Qwen3-VL-4B-Instruct + Full SFT 2026.05		58.2
InternVL-3.5 2026.03		57.3
Qwen3-VL + LOCUS 2026.06		56.7
Qwen3-VL-8B + Qwen3-8B / H-OPD 2026.07		55.9
Qwen3-VL 2026.03		55.3
Qwen3-VL-8B 2026.07		55.3
MiMo-VL + LOCUS 2026.06		54.8
Penguin-VL 2026.03		53.8
Step-GUI-8B 2025.12		53.7
GPT-4o 2026.03		53.2
Qwen3-VL 2026.06		52.5
MiMo-VL 2026.06		52.5
Qwen3-VL 8B-Instruct 2025.12		52.3
Step-GUI-4B 2025.12		52.1
LASER 2026.07		51.2
Kimi-vl-A3B-thinking 2026.05		51
VAPO-Thinker-7B 2026.07		50.9
PRPO 2026.06		49.7
VisionR1-7B 2026.07		49.7
Thyme 2026.03		49
DeepEyesV2 2026.03		48.7
VRE 2026.03		48.7
Qwen2.5-VL-7B + DUPL 2025.10		48.7
V-Zero 2026.01		48.6
Qwen2.5-VL-7B + PGPO 2026.04		47.93
Qwen2.5-VL-7B + DAPO 2026.04		47.85
Supervised GRPO 2026.01		47.8
InternVL3.5-2B 2025.12		47.7
DeepEyes 2025.05		47.7
InternVL3.5 2026.03		47.7
InternVL3.5 2026.03		47.7
Qwen2.5-VL-7B + VPPO 2026.04		47.52
NoisyRollout-7B 2026.06		47.5
VPPO 2026.06		47.5
Qwen2.5-VL-7B + NoisyRollout 2025.10		47.5
NoisyRollout-7B 2026.04		47.49
Qwen2.5-VL-7B + PAPO 2026.04		47.46
V-Zero 2026.01		47.4
VLAA-Thinker-7B 2025.10		47.3
MM-Eureka-Qwen-7B 2025.10		47.3
Qwen2.5-VL-7B-Instruct (Base) 2026.01		47.2
OpenVLThinker-7B 2025.10		47.1
PAPO-7B 2025.10		47.1
Qwen2.5-VL + LOCUS 2026.06		47.1
OST 2026.05		47
Qwen2.5-VL-7B + GRPO 2026.04		46.81
Qwen3-VL-2B-Instruct 2026.05		46.8
DEITA 2026.05		46.7
AutoTool 2026.05		46.7
PAPO-D 2026.06		46.7
Qwen3-VL-2B-Instruct + LLM-as-a-Judge 2026.05		46.5
VL-Rethinker-7B 2026.06		46.5
VL-Rethinker-7B 2026.04		46.47
MM-Eureka-7B 2026.04		46.3
MM-Eureka-7B 2026.06		46.3
InternVL3-8B 2026.05		46
Qwen2.5-VL-7B + GRPO 2025.10		46
Qwen2.5-VL* 2025.05		45.9
Qwen2.5-VL 2026.03		45.9
Qwen3-VL-2B-Instruct + Random 2026.05		45.9
R1-ShareVL-7B 2026.06		45.8
PAPO-G 2026.06		45.8
R1-ShareVL-7B 2026.04		45.76
Qwen3-VL-2B-Instruct + Full SFT 2026.05		45.6
GRPO 2026.06		45.6
R1-Onevision-7B 2026.07		45.6
Qwen2.5-VL-7B 2026.05		45.5
R1-Onevision-7B 2025.10		45.5
Qwen2.5-VL-7B 2025.10		45.5
DeepEyes 2026.05		45.3
OpenVLThinker-7B 2026.01		45.1
Qwen2.5-VL-3B + DUPL 2025.10		45
PRPO 2026.06		44.9
CFPOG 2026.06		44.45
BAGEL 2026.03		44.3
Qwen2.5-VL 2025.05		44.1
Qwen2.5-VL-3B + PGPO 2026.04		43.2
Qwen2.5-VL 2026.06		43.1
Qwen2.5-VL-7B 2026.04		43.01
Qwen2.5-VL-3B + PAPO 2026.04		42.7
Qwen2.5-VL-7B 2026.07		42.6

Showing 100 of 163 rows