VQA v2.0.

VQA-v2 test-dev. Qwen2-VL 72B. Table 1. arxiv:2409.12191

VQA-v2 test-dev. InternVL2-76B. Table 3. arxiv:2404.16821

VQA-v2 test-dev. InternVL2-76B. Table 3. arxiv:2404.16821

VQA-v2 test-dev. Table 5. Gemini 1.5 paper arxiv:2403.05530

VQA-v2 test-dev. Table 5. Gemini 1.5 paper arxiv:2403.05530

VQA-v2 test-dev. Table 5. Gemini 1.5 paper arxiv:2403.05530

VQA v2 test-dev. From Table 3 of PaLI-X paper (arxiv 2305.18565). State-of-the-art for encoder-decoder VLMs.

VQA v2 test-dev. From NVLM paper (arxiv 2409.11402) Table 7. Decoder-only architecture. Highest among open-access models at time of release.

VQA v2 test-dev. From NVLM paper (arxiv 2409.11402) Table 7. Cross-attention architecture.

VQA v2 test-dev. From NVLM paper (arxiv 2409.11402) Table 7. Hybrid architecture.

VQA v2 test-dev. Reported in NVLM paper (arxiv 2409.11402) Table 7. VILA-1.5 40B released Apr 2024.

VQA v2 test-dev. From official LLaVA-NeXT (LLaVA-1.6) blog post, Jan 2024. Best open-source at time of release.

VQA v2 test-dev. From official LLaVA-NeXT (LLaVA-1.6) blog post, Jan 2024.

CogVLM-17B. VQAv2 test-dev accuracy. NeurIPS 2024. Tsinghua/Zhipu.

VQA v2 test-dev. From official LLaVA-NeXT (LLaVA-1.6) blog post, Jan 2024.

VQA-v2 test-dev. FlanT5-XXL backbone. Table 9. arxiv:2301.12597

VQA-v2 test-dev. FlanT5-XXL backbone. Table 9. arxiv:2301.12597

VQA v2 test-dev. From official LLaVA-NeXT (LLaVA-1.6) blog post, Jan 2024.

VQA v2. Self-reported by Mistral AI. Pixtral Large 124B released Nov 2024. Score reported as 0.809 (80.9%).

VQA v2 test-dev. Reported in NVLM paper (arxiv 2409.11402) Table 7.

VQA-v2 test-dev. 13B (Vicuna) backbone. Table 1. arxiv:2310.03744

VQA-v2 test-dev. 13B (Vicuna) backbone. Table 1. arxiv:2310.03744

VQA v2 test-dev. From "Improved Baselines with Visual Instruction Tuning" (LLaVA-1.5), CVPR 2024. Also reported as baseline in LLaVA-NeXT blog.

VQA v2 test-dev. Reported in NVLM paper (arxiv 2409.11402) Table 7.

VQA v2. Self-reported by Mistral AI. Pixtral-12B released Sep 2024. Score reported as 0.786 (78.6%).

VQA-v2 test-dev. GPT-4o system card Table 1. arxiv:2410.21276

VQA-v2 test-dev. GPT-4o system card Table 1. arxiv:2410.21276

VQA v2. Reported by Meta for Llama 3.2 90B multimodal model. Self-reported score of 0.781 (78.1%).

VQA-v2 val, 0-shot. Table 2. GPT-4 Technical Report arxiv:2303.08774

VQA-v2 val, 0-shot. Table 2. GPT-4 Technical Report arxiv:2303.08774