AI2D

AI2D: Measures visual question answering, OCR, document understanding, chart comprehension, or layout-aware reasoning.

6rows
scoreprimary metric
2026-05-27sampled

Metadata

Metrics

AI2D

Latest Results

Rows are imported from direct AI2D metric values in the public ALL-Bench JSON leaderboard.

Rank Subject AI2D Model Match Provenance Sampled
1 Qwen3.5-9B 90.2 Qwen3.5-9B
qwen-qwen3.5-9b
Imported 2026-05-27
2 InternVL3-78B 89.7 Imported 2026-05-27
3 Qwen3.5-4B 89.6 Imported 2026-05-27
4 Qwen3-VL-30B-A3B 86.9 Imported 2026-05-27
5 Gemini-2.5-FL-Lite 85.7 Imported 2026-05-27
6 GPT-5-Nano 81.9 GPT-5 Nano
openai-gpt-5-nano
Imported 2026-05-27