Mathematical Reasoning in Vision on MathVista
74.7MathVista AccuracyVLsI-7B
Evaluation results
| Method | Links | ||
|---|---|---|---|
| VLsI-7BParameters=7B2024.12 | 74.7 | — | |
| BAGELType=Unified, LLM Parameters=7B MOT2025.05 | 73.1 | — | |
| Phantom-7BParameters=7B2024.12 | 70.9 | — | |
| Gemini-ProAccess Type=Closed-source API2024.07 | 70.2 | — | |
| Kimi-VLType=Und. Only, LLM Parameters=2.8B/16B2025.05 | 68.7 | — | |
| Qwen2.5-VLType=Und. Only, LLM Parameters=7B2025.05 | 68.2 | — | |
| InternVL2.5Type=Und. Only, LLM Parameters=7B2025.05 | 64.4 | — | |
| IXC-2.5-7BModel Size=7B2024.07 | 63.8 | — | |
| BAGELType=Unified, LLM Parameters=1.5B MOT2025.05 | 63.4 | — | |
| LLava-OVType=Und. Only, LLM Parameters=7B2025.05 | 63.2 | — | |
| LLaVA-OneVision-7BParameters=7B2024.12 | 63.2 | — | |
| DeepSeek-VL2Type=Und. Only, LLM Parameters=4.1B/28B2025.05 | 62.8 | — | |
| Qwen2.5-VLType=Und. Only, LLM Parameters=3B2025.05 | 62.3 | — | |
| MiniCPM-V2.6-8BParameters=8B2024.12 | 60.6 | — | |
| InternVL2Type=Und. Only, LLM Parameters=7B2025.05 | 58.3 | — | |
| InternVL2-8BParameters=8B2024.12 | 58.3 | — | |
| Qwen2-VLType=Und. Only, LLM Parameters=7B2025.05 | 58.2 | — | |
| Qwen2-VL-7BParameters=7B2024.12 | 58.2 | — | |
| Florence-VL 8B# Vis tok.=5762024.12 | 55.5 | — | |
| MUSE-VLType=Unified, LLM Parameters=32B2025.05 | 55 | — | |
| WeMMModel Size=8B, Access Type=Open-source2024.07 | 54.9 | — | |
| Eagle-13BParameters=13B2024.12 | 54.4 | — | |
| MiniCPM-V2.5-8BParameters=8B2024.12 | 54.3 | — | |
| Eagle-8BParameters=8B2024.12 | 52.7 | — | |
| Florence-VL 3B# Vis tok.=5762024.12 | 52.2 | — | |
| TroL-7BParameters=7B2024.12 | 51.8 | — | |
| InternVL2.5Type=Und. Only, LLM Parameters=1.8B2025.05 | 51.3 | — | |
| Cambrian-1-8BParameters=8B2024.12 | 49 | — | |
| Cambrian 8B# Vis tok.=5762024.12 | 49 | — | |
| Cambrian-1-13BParameters=13B2024.12 | 48 | — | |
| GPT-4VAccess Type=Closed-source API2024.07 | 47.8 | — | |
| InternVL2Type=Und. Only, LLM Parameters=1.8B2025.05 | 46.4 | — | |
| Qwen2-VLType=Und. Only, LLM Parameters=1.5B2025.05 | 43 | — | |
| MM1-MoE-7B×32Parameters=7B×322024.12 | 40.9 | — | |
| BLIP-3Type=Und. Only, LLM Parameters=4B2025.05 | 39.6 | — | |
| LATTETraining Data Size=293K2024.12 | 38.9 | — | |
| LLaVA-NeXT-8BParameters=8B2024.12 | 37.5 | — | |
| LLaVA next 8B# Vis tok.=28802024.12 | 37.4 | — | |
| MiniGemini-HD-13BParameters=13B2024.12 | 37 | — | |
| Mini-Gemini-HD 8B# Vis tok.=28802024.12 | 37 | — | |
| LATTETraining Data Size=98K2024.12 | 36.9 | — | |
| MM1-7BParameters=7B2024.12 | 35.9 | — | |
| CogCoMTraining Data Size=70K2024.12 | 35.7 | — | |
| LLaVA-NeXT-13BParameters=13B2024.12 | 35.1 | — | |
| LLaVA-NeXT-7BParameters=7B2024.12 | 34.6 | — | |
| MiniGemini-HD-7BParameters=7B2024.12 | 32.2 | — | |
| Vila 3B2024.12 | 30.6 | — | |
| ChameleonType=Unified, LLM Parameters=7B2025.05 | 22 | — | |
| IDEFICS-80BLLM Backbone=LLaMA-65B2023.12 | — | 26.2 | |
| InstructBLIPLLM Backbone=Vicuna-7B2023.12 | — | 25.3 | |
| LLaVALLM Backbone=Vicuna-7B2023.12 | — | 23.7 | |
| LLaVA-1.5LLM Backbone=Vicuna-13B2023.12 | — | 26.1 | |
| LyricsLLM Backbone=Vicuna-13B2023.12 | — | 39.4 | |
| Qwen-VL-ChatLLM Backbone=Qwen-7B2023.12 | — | 33.8 | |
| ShareGPT4VLLM Backbone=Vicuna-7B2023.12 | — | 25.8 |