Multimodal benchmark
VPCT leaderboard
As of October 2026, Gemini 3 Pro has the highest published VPCT score on Noometry at 91%, out of 24 models with results.
Last verified
About VPCT
Visual physics comprehension: predict where a ball will land in a diagram.
- Category
- Multimodal
- Introduced
- 2025
- Format
- Multiple choice
- Unit
- Percent (random guessing ≈ 33.3%)
- Official site
- cbrower.dev
Top 15 models
- Gemini 3 Pro 91%
- GPT-5.2 84%
- Gemini 3 Flash Preview 72.6%
- GPT-5 66%
- GPT-5.1 58.7%
- o4-mini 57.5%
- o3 52%
- Gemini 2.5 Pro 48%
- Gemini 2.5 Flash 46.2%
- GPT-4.5 45%
- GPT-5 Mini 40.2%
- Claude Opus 4.5 40%
- GPT-4o 40%
- Claude Sonnet 4.5 39.8%
- Claude 3.7 Sonnet 39%
Sponsored placements are available on pages like this one. Advertise on Noometry
All results
| # | Model | Provider | Score | Setting | Source | Date |
|---|---|---|---|---|---|---|
| 1 | Gemini 3 Pro |  Google | 91% | Epoch AI | ||
| 2 | GPT-5.2 | OpenAI | 84% | xhigh | Epoch AI | |
| 3 | Gemini 3 Flash Preview | Google | 72.6% | Epoch AI | ||
| 4 | GPT-5 | OpenAI | 66% | high | Epoch AI | |
| 5 | GPT-5.1 | OpenAI | 58.7% | high | Epoch AI | |
| 6 | o4-mini | OpenAI | 57.5% | medium | Epoch AI | |
| 7 | o3 | OpenAI | 52% | medium | Epoch AI | |
| 8 | Gemini 2.5 Pro | Google | 48% | Epoch AI | ||
| 9 | Gemini 2.5 Flash | Google | 46.2% | Epoch AI | ||
| 10 | GPT-4.5 | OpenAI | 45% | Epoch AI | ||
| 11 | GPT-5 Mini | OpenAI | 40.2% | medium | Epoch AI | |
| 12 | Claude Opus 4.5 | Anthropic | 40% | 32K | Epoch AI | |
| 13 | GPT-4o | OpenAI | 40% | Epoch AI | ||
| 14 | Claude Sonnet 4.5 | Anthropic | 39.8% | 32K | Epoch AI | |
| 15 | Claude 3.7 Sonnet | Anthropic | 39% | Epoch AI | ||
| 16 | Claude Opus 4 | Anthropic | 38% | 16K | Epoch AI | |
| 17 | GPT-5 Nano | OpenAI | 37.2% | high | Epoch AI | |
| 18 | o1 | OpenAI | 37% | medium | Epoch AI | |
| 19 | Claude Opus 4.1 | Anthropic | 35% | Epoch AI | ||
| 20 | Claude Sonnet 4 | Anthropic | 34% | 32K | Epoch AI | |
| 21 | GPT-4o mini | OpenAI | 34% | Epoch AI | ||
| 22 | Claude 3.5 Sonnet | Anthropic | 33% | Epoch AI | ||
| 23 | Claude 3.5 Sonnet | Anthropic | 33% | Epoch AI | ||
| 24 | Gemini 2.5 Flash-Lite | Google | 30% | Epoch AI |
Compare the leaders
Other multimodal benchmarks
- LMArena Vision
- Video-MME
- GeoBench
- Blueprint-Bench 2
- Furniture Assembly
- LMArena Document (reference)
- MindCube (reference)
- ScienceQA (reference)
- SpatialViz-Bench (reference)
Frequently asked questions
What does VPCT measure?
Visual physics comprehension: predict where a ball will land in a diagram.
Which model has the highest VPCT score?
As of October 2026, Gemini 3 Pro has the highest published VPCT score on Noometry at 91%, out of 24 models with results.