AI2D_TEST
AI2D_TEST - grade-school science diagram understanding benchmark.
Category
Document Understanding
Max Score
100.0
Score Type
percent
Active
Yes
Model Scores
Scores for AI2D_TEST
| Model | Score | Date | Verified |
|---|---|---|---|
| Gemma4-31B-it |
85.71%
|
16.04.2026 | Unverified |
| Gemma4-26B-A4B |
83.39%
|
16.04.2026 | Unverified |
| Qwen3.5-27B |
98.67%
|
16.04.2026 | Unverified |
| Gemma 3 4B PT |
63.20
|
27.02.2025 | Verified |
| Gemma 3 12B PT |
39.87%
|
27.02.2025 | Verified |
| Gemma 3 27B PT |
52.49%
|
27.02.2025 | Verified |
| Qwen3.6-35B-A3B |
98.01%
|
16.04.2026 | Verified |
| Qwen3.5-35B-A3B |
97.67%
|
16.04.2026 | Unverified |
| Claude-Sonnet-4.5 |
79.07%
|
16.04.2026 | Unverified |
| Qwen3.5-4B |
87.71%
|
01.02.2026 | Verified |
| Qwen3.5-122B-A10B |
100.00%
|
01.02.2026 | Verified |
| GPT-5-mini 2025-08-07 |
83.06%
|
01.02.2026 | Unverified |
| Qwen3-235B-A22B |
86.38%
|
01.02.2026 | Unverified |
| Qwen3-VL-235B-A22B |
86.38%
|
01.02.2026 | Verified |