List your company
ResearchFormerly WizWand

Mathematical Reasoning on MathVista

229.2ScoreQwen-VL-7B-Chat
63.008106.154149.3192.446Nov 13, 2023Apr 18, 2024Sep 23, 2024Feb 28, 2025Aug 5, 2025Jan 10, 2026Jun 17, 2026
Updated 4mo ago

Evaluation results

MethodLinks
Qwen-VL-7B-ChatModel Size=7B, Mode=Chat2023.11229.2
EcoAlignCategory=Closed-source, Model=Qwen-VL-Max, Avg. Cost=12.72025.1190.7
EcoAlignCategory=Closed-source, Model=Gemini-2.5-Flash, Avg. Cost=24.12025.1189.6
CoTCategory=Closed-source, Model=Qwen-VL-Max, Avg. Cost=1142025.1189.5
CoDCategory=Closed-source, Model=Qwen-VL-Max, Avg. Cost=34.32025.1189.1
CoDCategory=Closed-source, Model=Gemini-2.5-Flash, Avg. Cost=27.52025.1188.5
CoTCategory=Closed-source, Model=Gemini-2.5-Flash, Avg. Cost=86.62025.1188.2
CoTCategory=Closed-source, Model=GPT-4o, Avg. Cost=104.32025.1186.8
EcoAlignCategory=Open-source, Model=InternVL3-14B, Avg. Cost=39.32025.1186
EcoAlignCategory=Closed-source, Model=GPT-4o, Avg. Cost=21.22025.1185.4
CoDCategory=Closed-source, Model=GPT-4o, Avg. Cost=25.82025.1184.2
CoDCategory=Open-source, Model=InternVL3-14B, Avg. Cost=45.52025.1184.1
CoTCategory=Open-source, Model=InternVL3-14B, Avg. Cost=134.22025.1183.9
LongCat-Next2026.0383.1
Gemini-2.5-ProDate=2024.12, Tokenization Paradigm=Proprietary Models2026.0282.7
Gemini-2.5-ProModel Category=Proprietary Models2026.0682.7
GPT-5Inference mode=Thinking mode2026.0481.9
MiMo-VL-RLLLM Backbone=MiMo-7B, Open Data=false2025.1281.5
Qwen3-VLInference mode=Thinking mode, Size=8B2026.0481.4
GPT-5-highModel Category=Proprietary Models2026.0681.3
MiniCPM-o 4.5Inference mode=Thinking mode, Size=9B2026.0481
VL-Scaler-MiMOSize=7B2026.0580.3
MiniCPM-o 4.5Size=9B, mode=instruct2026.0480.1
Qwen3-OmniInference mode=Thinking mode, Size=30B-A3B2026.0480
MetaForgeTool Setting=w/ OOD Tools2026.0679.78
Gemini 2.5 FlashInference mode=Thinking mode2026.0479.4
BaseCategory=Closed-source, Model=Qwen-VL-Max, Avg. Cost=12025.1179
MiMO-VL-InstructSize=7B2026.0578.8
InternVL3.5Size=8B, mode=instruct2026.0478.4
OPDModel Scale=8B2026.0678.1
MetaForgeTool Setting=w/ IID Tools2026.0678.01
BaseCategory=Closed-source, Model=Gemini-2.5-Flash, Avg. Cost=12025.1177.7
Qwen3-VLSize=8B, mode=instruct2026.0477.2
InternVL3.5Model Scale=4B, Date=2025.08, Tokenization Paradigm=Understanding-only Models2026.0277.1
OPD+ViCuRModel Scale=8B2026.0676.7
GRPOModel Scale=8B2026.0676.6
TVI-CoTModel Category=MLLM-based Chain-of-Thought Methods, Model Scale=8B2026.0676.6
KelixModel Scale=8B, Tokenization Paradigm=Discrete Tokenization2026.0276.5
LVRPOType=Unified (>1.5B), # LLM Params=7B2026.0376.2
VLM-GuardCategory=Closed-source, Model=Qwen-VL-Max, Avg. Cost=2.22025.1176.1
Qwen3-OmniSize=30B-A3B, mode=instruct2026.0475.9
Base ModelModel Scale=8B2026.0675.8
VAPO-Thinker-7BModel Category=MLLM-based Chain-of-Thought Methods, Model Scale=7B2026.0675.6
Gemini 2.5 FlashSize=-, mode=instruct2026.0475.3
BiPS-General-7BData=13K+39K2025.1275
VL-RethinkerSize=7B2026.0574.9
VL-Rethinker-7BModel Category=MLLM-based Chain-of-Thought Methods, Model Scale=7B2026.0674.9
Qwen2.5-VL-InstructSize=72B2026.0574.8
Qwen3-VL-4B (Base)Base Model=Qwen3-VL-4B, Alignment Strategy=Base2026.0674.8
VLsIParameters=7B2024.1274.7
VLSI-7BSize=7B2024.1274.7
Claude-3.7-SonnetData=-2025.1274.5
Claude-Opus-4.1Model Category=Proprietary Models2026.0674.5
GRPOData=13K + 39K2025.1274.3
OPSD+ViCuRModel Scale=8B2026.0674.1
VL-ScalerSize=7B2026.0573.8
BiPS-Chart-7BData=13K2025.1273.5
BaseCategory=Open-source, Model=InternVL3-14B, Avg. Cost=12025.1173.5
OPSDModel Scale=8B2026.0673.5
Vision-R1-7BModel Category=MLLM-based Chain-of-Thought Methods, Model Scale=7B2026.0673.5
ManzanoModel Scale=30B, Date=2025.09, Tokenization Paradigm=Hybrid Tokenization2026.0273.3
Vision-R1-7BData=73K + 137K2025.1273.2
Qwen3-VL-8B (Baseline)Model Category=Baseline, Model Scale=8B2026.0673.2
BagelModel Scale=14B, Date=2025.05, Tokenization Paradigm=Continuous Tokenization2026.0273.1
Bagel#LLM=14B2025.1273.1
BaseCategory=Closed-source, Model=GPT-4o, Avg. Cost=12025.1173.1
VLM-GuardCategory=Closed-source, Model=Gemini-2.5-Flash, Avg. Cost=1.42025.1173.1
BAGEL2026.0373.1
BAGELType=Unified (>1.5B), # LLM Params=7B2026.0373.1
BAGALType=Und., # Params=14B2025.0373.1
COPSD (Standard)Base Model=Qwen3-VL-4B, Alignment Strategy=COPSD (Standard)2026.0672.7
COPSD (Hybrid)Base Model=Qwen3-VL-4B, Alignment Strategy=COPSD (Hybrid)2026.0672.5
PRPOModel Scale=7B, Base Model=Qwen2.5-VL-7B, Rollout=82026.0672.2
Qwen3-VLScale=4B2026.0572
MM-Eureka-7BModel Scale=7B, Rollout=82026.0671.9
InternVL3.5Number of Parameters=2.3B2025.1271.8
CodeV-7B-RLTool use=Yes, Training=RL2025.1171.8
InternVL3LLM Backbone=Qwen2.5-7B, Open Data=false2025.1271.6
InternVL3-8BModel Category=Open-Source MLLMs, Model Scale=8B2026.0671.6
InternVL3Model Size=8B2026.0671.6
Qwen3.5-VLScale=2B2026.0571.4
ThinkLite-7BTool use=No2025.1171.3
Pixel-Reasoner-7BTool use=Yes2025.1171.2
SAIL-VL2Number of Parameters=2.7B2025.1271.1
Phantom-7BSize=7B2024.1270.9
PhantomParameters=7B2024.0970.9
DeepEyes-7BData=14K + 33K2025.1270.8
VL-Rethinker-7BModel Scale=7B, Rollout=82026.0670.6
VLM-GuardCategory=Open-source, Model=InternVL3-14B, Avg. Cost=1.92025.1170.5
VPPOModel Scale=7B, Base Model=Qwen2.5-VL-7B, Rollout=82026.0670.5
InternVL3-8BData=-2025.1270.4
InternVL3-8BTool use=No2025.1170.4
Latent DenoisingArchitecture=Qwen-2.5-VL2026.0470.3
OpenVLThinkerSize=7B2026.0570.2
Qwen2-VL-72BSize=72B2024.1269.7
VLM-GuardCategory=Closed-source, Model=GPT-4o, Avg. Cost=3.12025.1169.7
LLaVA-OneVision-1.5-8BModel Category=Open-Source MLLMs, Model Scale=8B2026.0669.6
Ovis-U1#LLM=1.5B2025.1269.4
Ovis-U12026.0369.4
SFTBase Model=Qwen3-VL-4B, Alignment Strategy=SFT2026.0669.4