multimodal benchmark
AI2D Leaderboard
AI2D is a multimodal benchmark based on 4,903 grade school natural science diagrams and over 15,000 multiple-choice questions and answers.
Updated Sep 7, 2026
Models34
Model coverage34
MetricScore
EvidenceB
AI2D Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelAN | Score94.70% | Percentile100.00% | Participants34 | EvidenceC | Evaluated |
| Rank02 | ModelAC | Score94.40% | Percentile96.97% | Participants34 | EvidenceC | Evaluated |
| Rank03 | ModelOP | Score94.20% | Percentile93.94% | Participants34 | EvidenceC | Evaluated |
| Rank04 | ModelMA | Score93.80% | Percentile90.91% | Participants34 | EvidenceC | Evaluated |
| Rank05 | ModelAC | Score93.30% | Percentile87.88% | Participants34 | EvidenceC | Evaluated |
| Rank06 | ModelMA | Score92.91% | Percentile84.85% | Participants34 | EvidenceC | Evaluated |
| Rank07 | ModelAC | Score92.90% | Percentile81.82% | Participants34 | EvidenceC | Evaluated |
| Rank08 | ModelAC | Score92.70% | Percentile78.79% | Participants34 | EvidenceC | Evaluated |
| Rank09 | ModelAC | Score92.60% | Percentile75.76% | Participants34 | EvidenceC | Evaluated |
| Rank10 | ModelME | Score92.30% | Percentile72.73% | Participants34 | EvidenceC | Evaluated |
| Rank11 | ModelME | Score91.10% | Percentile69.70% | Participants34 | EvidenceC | Evaluated |
| Rank12 | ModelAC | Score89.70% | Percentile66.67% | Participants34 | EvidenceC | Evaluated |
| Rank13 | ModelAC | Score89.50% | Percentile63.64% | Participants34 | EvidenceC | Evaluated |
| Rank14 | ModelAC | Score89.20% | Percentile60.61% | Participants34 | EvidenceC | Evaluated |
| Rank15 | ModelLA | Score89.00% | Percentile57.58% | Participants34 | EvidenceC | Evaluated |
| Rank16 | ModelAC | Score88.90% | Percentile54.55% | Participants34 | EvidenceC | Evaluated |
| Rank17 | ModelAC | Score88.40% | Percentile51.52% | Participants34 | EvidenceC | Evaluated |
| Rank18 | ModelXA | Score88.30% | Percentile48.48% | Participants34 | EvidenceC | Evaluated |
| Rank19 | ModelAC | Score86.90% | Percentile45.45% | Participants34 | EvidenceC | Evaluated |
| Rank20 | ModelAC | Score85.70% | Percentile42.42% | Participants34 | EvidenceC | Evaluated |
| Rank21 | ModelAC | Score85.00% | Percentile39.39% | Participants34 | EvidenceC | Evaluated |
| Rank22 | ModelAC | Score84.90% | Percentile36.36% | Participants34 | EvidenceC | Evaluated |
| Rank23 | ModelAC | Score84.90% | Percentile33.33% | Participants34 | EvidenceC | Evaluated |
| Rank24 | ModelGO | Score84.50% | Percentile30.30% | Participants34 | EvidenceC | Evaluated |
| Rank25 | ModelGO | Score84.20% | Percentile27.27% | Participants34 | EvidenceC | Evaluated |
| Rank26 | ModelAC | Score84.10% | Percentile24.24% | Participants34 | EvidenceC | Evaluated |
| Rank27 | ModelAC | Score83.20% | Percentile21.21% | Participants34 | EvidenceC | Evaluated |
| Rank28 | ModelMI | Score82.30% | Percentile18.18% | Participants34 | EvidenceC | Evaluated |
| Rank29 | ModelDE | Score81.40% | Percentile15.15% | Participants34 | EvidenceC | Evaluated |
| Rank30 | ModelDE | Score80.00% | Percentile12.12% | Participants34 | EvidenceC | Evaluated |
AI2D Highlights
The leading models and scores on this benchmark.
AI2D Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for AI2D
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is AI2D?
What AI2D measures and how its scores work.
AI2D is a dataset and benchmark featuring illustrative diagrams from grade school natural sciences, including food webs, human physiology, and life cycles.
It evaluates diagram understanding and visual reasoning by assessing how models interpret diagrammatic elements, relationships, and structure to answer questions about scientific concepts represented visually, using Score reported as a ratio.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- AI2D
- Modality
- multimodal
- Primary category
- multimodal
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about AI2D.
Which model scores highest on AI2D?
Claude 3.5 Sonnet is currently ranked first with 94.70%.
What are the top three models on AI2D?
The current leaders are Claude 3.5 Sonnet (94.70%), Qwen3.6 Plus (94.40%), and GPT-4o (94.20%).
Which AI2D model has the lowest official input price?
Qwen2.5-Omni-7B has the lowest matched official input price at $0.10 input / $0.40 output per 1M tokens.
Which models are fastest among AI2D results?
The fastest matched records are GPT-4o (132.00 tok/s via OpenAI), Llama 3.2 11B Instruct (108.00 tok/s via DeepInfra), and Claude 3.5 Sonnet (100.00 tok/s via Anthropic).
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.