Chart Visual Question Answering on CHARTOM misleading visualizations
55.9QA AccuracyGPT5
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| GPT52025.02 | 55.9 | — | |
| GPT5-mini2025.02 | 53.2 | — | |
| Gemini-2.5-flash2025.02 | 42 | — | |
| claude-3.5-sonnet2025.02 | 39.9 | 31.5 | |
| Claude-3.5-sonnet2025.02 | 39.9 | — | |
| gpt4o2025.02 | 36.4 | 28.7 | |
| gemini-1.5-pro2025.02 | 34.3 | 26.6 | |
| internvl2.5-38BModel size=38B2025.02 | 31.5 | 23.8 | |
| gpt42025.02 | 30.8 | 23.1 | |
| chartinstruction-13BModel size=13B2025.02 | 28 | 21 | |
| gemini-1.5-flash2025.02 | 28 | 21 | |
| qwen2vl-2BModel size=2B2025.02 | 27.3 | 20.3 | |
| ovis1.6-27BModel size=27B2025.02 | 26.6 | 19.6 | |
| llava-13BModel size=13B2025.02 | 25.9 | 18.9 | |
| llava-7BModel size=7B2025.02 | 25.2 | 18.2 | |
| ovis1.6-9BModel size=9B2025.02 | 24.5 | 17.5 | |
| internvl2.5-8BModel size=8B2025.02 | 24.5 | 17.5 | |
| qwen2vl-7BModel size=7B2025.02 | 22.4 | 15.4 | |
| internvl2.5-26BModel size=26B2025.02 | 21.7 | 15.4 | |
| internvl2.5-4BModel size=4B2025.02 | 21 | 14.7 | |
| tinychart-3BModel size=3B2025.02 | 18.9 | 12.6 | |
| chartgemma-3BModel size=3B2025.02 | 18.9 | 12.6 | |
| internvl2.5-2BModel size=2B2025.02 | 16.1 | 10.5 |