Table Question Answering on TABMWP
94.97AccuracyQwen3-VL-8B + TABQAWORLD
Evaluation results
| Method | Links | ||
|---|---|---|---|
| Qwen3-VL-8B + TABQAWORLDStrategy=Training-free Agent2026.04 | 94.97 | — | |
| DiVA-FormerModality=V+T2026.03 | 92 | — | |
| SynTab-LLaVA-7BModality=Image2026.04 | 88.3 | — | |
| MiniCPM-V-2.6 + HIPPOScale=8B, Modality=Image & Text2025.02 | 87.5 | 65.46 | |
| HIPPO-8BModality=Multimodal2026.04 | 87.5 | — | |
| Ovis2-8B (Image-only Path)Strategy=Dynamic Adaptive Routing2026.04 | 87 | — | |
| Qwen2.5-VL-7B + TABQAWORLDStrategy=Training-free Agent2026.04 | 86.43 | — | |
| MiniCPM-V-2.6Scale=8B, Modality=Image & Text2025.02 | 86.06 | 62.12 | |
| TableDART (TG2-7B+Ovis2-8B)Strategy=Dynamic Adaptive Routing2026.04 | 84.54 | — | |
| TableGPT2-7B (Text-only Path)Strategy=Dynamic Adaptive Routing2026.04 | 83.87 | — | |
| MiniCPM-V-2.6 (Image Only)Scale=8B, Modality=Image2025.02 | 83.68 | 60.42 | |
| MiniCPM-V-2.6-8BModality=Image2026.04 | 83.68 | — | |
| AdapterModality=V+T2026.03 | 82.8 | — | |
| Qwen2.5-VL-Instruct + HIPPOScale=7B, Modality=Image & Text2025.02 | 82.55 | 62.9 | |
| AdapterModality=Vision2026.03 | 82 | — | |
| Qwen3-VL-8BModality=Image2026.04 | 79.9 | — | |
| Qwen2.5-InstructScale=7B, Modality=Text2025.02 | 79.45 | 61.54 | |
| Table-LLaVAScale=7B, Modality=Image & Text2025.02 | 78.7 | 50.98 | |
| AdapterModality=Text2026.03 | 78.4 | — | |
| Table-LLaVA + HIPPOScale=7B, Modality=Image & Text2025.02 | 78.22 | 55.52 | |
| Qwen2-72B-InstructScale=72B, Modality=Text2025.02 | 77.49 | 57.11 | |
| Qwen2.5-VL-Instruct + Vanilla SFTScale=7B, Modality=Image & Text2025.02 | 77.17 | 62.15 | |
| MiniCPM-V-2.6 + Vanilla SFTScale=8B, Modality=Image & Text2025.02 | 76.69 | 61.93 | |
| DirectModality=Text2026.03 | 75.9 | — | |
| DirectModality=Vision2026.03 | 74.1 | — | |
| DirectModality=V+T2026.03 | 73.8 | — | |
| TableDART (TG2-7B+Qwen2.5-VL-7B)Strategy=Dynamic Adaptive Routing2026.04 | 72.61 | — | |
| Qwen2.5-VL-Instruct (Image Only)Scale=7B, Modality=Image2025.02 | 65.63 | 56.85 | |
| GPT-4oScale=N/A, Modality=Image & Text2025.02 | 64.14 | 55.58 | |
| AdaCADModel=Ministral-3-8B2026.04 | 63.7 | — | |
| Qwen2.5-VL-7BModality=Image2026.04 | 63.69 | — | |
| NWCADModel=Ministral-3-8B2026.04 | 63.6 | — | |
| With-contextModel=Ministral-3-8B2026.04 | 63.4 | — | |
| GPT-4o (Image Only)Scale=N/A, Modality=Image2025.02 | 63.27 | 54.26 | |
| CADModel=Ministral-3-8B2026.04 | 62.7 | — | |
| Qwen2.5-VL-InstructScale=7B, Modality=Image & Text2025.02 | 62.59 | 58.58 | |
| NWCADModel=Llama-3.1-70B2026.04 | 61 | — | |
| AdaCADModel=Llama-3.1-70B2026.04 | 60.12 | — | |
| With-contextModel=Llama-3.1-70B2026.04 | 60 | — | |
| Table-LLaVA (Image Only)Scale=13B, Modality=Image2025.02 | 59.77 | 38.09 | |
| Table-LLaVA (Image Only)Scale=7B, Modality=Image2025.02 | 57.78 | 35.69 | |
| Table-LLaVA-7BModality=Image2026.04 | 57.78 | — | |
| CoCoAModel=Llama-3.1-70B2026.04 | 57.62 | — | |
| TableDreamerIFT Data=27K, Backbone=GPT-4o2025.06 | 54.66 | — | |
| With-contextModel=Llama-3.1-8B2026.04 | 54.4 | — | |
| AdaCADModel=Llama-3.1-8B2026.04 | 54.4 | — | |
| NWCADModel=Llama-3.1-8B2026.04 | 54.4 | — | |
| CoCoAModel=Ministral-3-8B2026.04 | 53.4 | — | |
| Llama3.1-InstructScale=8B, Modality=Text2025.02 | 51.86 | 35.42 | |
| CoCoAModel=Llama-3.1-8B2026.04 | 50.6 | — | |
| CADModel=Llama-3.1-70B2026.04 | 49.88 | — | |
| TableGPT2IFT Data=2.36M, Backbone=Qwen2.5-7B2025.06 | 49.35 | — | |
| CADModel=Llama-3.1-8B2026.04 | 49.1 | — | |
| Gemini 2.0 FlashModality=Multimodal2026.04 | 46.29 | — | |
| TableDreamerIFT Data=27K, Backbone=Llama3.1-70B-Instruct2025.06 | 42.47 | — | |
| TableLLM-syn-dataIFT Data=80K2025.06 | 42.24 | — | |
| Llama3-InstructScale=8B, Modality=Text2025.02 | 42.01 | 31.98 | |
| Llama3-Instruct-8BModality=Text2026.04 | 42.01 | — | |
| GenQAIFT Data=100K2025.06 | 41.06 | — | |
| Yi-1.5-9B-Chat2025.06 | 38.23 | — | |
| TableLLMIFT Data=80K, Backbone=CodeLlama-7B2025.06 | 37.86 | — | |
| Llama3.1-8B-Instruct2025.06 | 36.53 | — | |
| MagpieIFT Data=100K2025.06 | 34.66 | — | |
| MiniCPM3-4B2025.06 | 34.06 | — | |
| InternLM2.5-7B-Chat2025.06 | 32.59 | — | |
| Evol-InstructIFT Data=100K2025.06 | 31.83 | — | |
| Self-InstructIFT Data=100K2025.06 | 28.98 | — | |
| DynasourIFT Data=132K2025.06 | 28.59 | — | |
| Mistral-7B-Instruct-v0.32025.06 | 25.71 | — | |
| Llama2Scale=7B, Modality=Text2025.02 | 22.82 | 17.53 | |
| Llama-2-7BModality=Text2026.04 | 22.82 | — | |
| OmniTabIFT Data=100K2025.06 | 22.67 | — | |
| GLM4-9B-Chat2025.06 | 20.3 | — | |
| ReasTapIFT Data=100K2025.06 | 19.54 | — | |
| Phi3.5-mini-3.8B2025.06 | 19.26 | — | |
| TabPediaScale=7B, Modality=Image2025.02 | 18.88 | 20.84 | |
| TableBenchLLMIFT Data=20K, Backbone=Llama3.1-8B2025.06 | 18.5 | — | |
| ResamplerModality=V+T2026.03 | 18.4 | — | |
| ResamplerModality=Vision2026.03 | 18 | — | |
| ResamplerModality=Text2026.03 | 17.2 | — | |
| TableGPT-syn-dataIFT Data=66K2025.06 | 16.13 | — | |
| DeepSeek-V2-Lite-16B-Chat2025.06 | 15.65 | — | |
| BaselineModel=Ministral-3-8B2026.04 | 13.6 | — | |
| MonkeyScale=7B, Modality=Image2025.02 | 13.26 | 14.16 | |
| BaselineModel=Llama-3.1-70B2026.04 | 12.62 | — | |
| UCTRIFT Data=43K2025.06 | 12.03 | — | |
| BaselineModel=Llama-3.1-8B2026.04 | 11.9 | — | |
| TableLlama + MarkdownScale=7B, Modality=Text2025.02 | 10.1 | 37.86 | |
| TableLlama-7BModality=Text2026.04 | 10.1 | — | |
| mPLUG-Owl2Scale=7B, Modality=Image2025.02 | 6.83 | 7.76 | |
| LLaVA v1.5Scale=7B, Modality=Image2025.02 | 6.05 | 9.82 | |
| Baichuan2-7B-Chat2025.06 | 4.6 | — | |
| Vary-toyScale=1.8B, Modality=Image2025.02 | 4.42 | 5.77 | |
| Qwen-VLScale=7B, Modality=Image2025.02 | 3.3 | 0.83 | |
| mPLUG-OwlScale=7B, Modality=Image2025.02 | 1.76 | 3.16 | |
| MiniGPT-4Scale=7B, Modality=Image2025.02 | 0.22 | 0.28 | |
| InternLM-XComposerScale=7B, Modality=Image2025.02 | 0.06 | 0.77 | |
| TableLlama + OracleScale=7B, Modality=Text2025.02 | — | 55.33 |