Visual Question Answering on CountQA
66.88Accuracy (%)Gemini-3-Flash
Evaluation Results
| Method | Links | |
|---|---|---|
| Gemini-3-Flash2026.02 | 66.88 | |
| Kimi-K2.5 (1T)number of parameters=1T2026.02 | 52.81 | |
| Qwen3-VL-235B-A22Bnumber of parameters=235B2026.02 | 40.58 | |
| GLM-4.5V (108B)number of parameters=108B2026.02 | 35.93 | |
| ZwZ-8Bnumber of parameters=8B2026.02 | 32.4 | |
| GPT-5.12026.02 | 31.41 | |
| ZwZ-4Bnumber of parameters=4B2026.02 | 30.82 | |
| Qwen3-VL-8Bnumber of parameters=8B2026.02 | 28.99 | |
| MiMo-VL-7B-RLnumber of parameters=7B2026.02 | 28.27 | |
| Qwen3-VL-4Bnumber of parameters=4B2026.02 | 28.14 | |
| MiniCPM-V-4.5 (9B)number of parameters=9B2026.02 | 23.43 | |
| ZwZ-7Bnumber of parameters=7B2026.02 | 20.72 | |
| Qwen2.5-VL-7Bnumber of parameters=7B2026.02 | 18.91 |