VPCT leaderboard
As of 6 Nov 2026, the top score on VPCT is 91.0% by Gemini 3 Pro (Google), ahead of GPT-5.2 at 84.0%, across 30 model configurations with a published result (Epoch AI Benchmarking Hub, result dated 18 Nov 2025). Images of physical setups; the model predicts the outcome. VPCT is shown for reference only and does not count toward the BenchLeader Index.
Key facts (as of 6 Nov 2026)
- Gemini 3 Pro holds the top VPCT score at 91.0% (Epoch AI Benchmarking Hub, 18 Nov 2025).
- GPT-5.2 is second on VPCT at 84.0% (Epoch AI Benchmarking Hub, 11 Dec 2025).
- VPCT: Visual physics comprehension: predicting what happens next in a scene (Epoch AI Benchmarking Hub).
- 30 model configurations have a published VPCT result on BenchLeader (BenchLeader, 6 Nov 2026).
- VPCT does not feed the BenchLeader Index; it is shown for reference (BenchLeader methodology).
Visual physics comprehension: predicting what happens next in a scene.
- Published by
- Epoch AI Benchmarking Hub
- Category
- Multimodal
- Index weight
- Reference only
- Models
- 30
- Data as of
- 6 Nov 2026
CC BY 4.0 — Epoch AI, ‘AI Benchmarking Hub’, epoch.ai/benchmarks. Mirrored boards credit their original publishers.
What the test looks like
Images of physical setups; the model predicts the outcome.
How it is scored
Percent correct, as published.
What to keep in mind
Vision models only; small set.
- 1Gemini 3 Pro91.0%
- 2GPT-5.2 (xhigh)84.0%
- 3Gemini 3 Flash72.6%
- 4GPT-5.2 (high)67.0%
- 5GPT-5 (high)66.0%
- 6GPT-5 (medium)63.2%
- 7GPT-5.1 (high)58.7%
- 8o4-mini (medium)57.5%
- 9GPT-5.1 (medium)53.3%
- 10o3 (medium)52.0%
- 11Gemini 2.5 Pro48.0%
- 12Gemini 2.5 Flash 09 202546.2%
- 13GPT-4.545.0%
- 14gemini-robotics-er-1.5-preview40.8%
- 15GPT-5 mini (medium)40.2%
| # | ||||
|---|---|---|---|---|
| 1 |  Gemini 3 Pro | 91.0% | 60.2 | 2025-11-18 |
| 2 |  GPT-5.2xhigh | 84.0% | 60.0 | 2025-12-11 |
| 3 | Gemini 3 Flash | 72.6% | 56.9 | 2025-12-17 |
| 4 | GPT-5.2high | 67.0% | 56.0 | 2025-12-11 |
| 5 | GPT-5high | 66.0% | 57.9 | 2025-08-07 |
| 6 | GPT-5medium | 63.2% | 58.1 | 2025-08-07 |
| 7 | GPT-5.1high | 58.7% | 57.7 | 2025-11-13 |
| 8 | o4-minimedium | 57.5% | 51.8 | 2025-04-16 |
| 9 | GPT-5.1medium | 53.3% | 52.2 | 2025-11-13 |
| 10 | o3medium | 52.0% | 53.3 | 2025-04-16 |
| 11 | Gemini 2.5 Pro | 48.0% | 54.5 | 2025-03-31 |
| 12 | Gemini 2.5 Flash 09 2025 | 46.2% | 46.8 | 2025-09-25 |
| 13 | GPT-4.5 | 45.0% | 50.8 | 2025-02-27 |
| 14 | gemini-robotics-er-1.5-preview | 40.8% | – | 2025-09-26 |
| 15 | GPT-5 minimedium | 40.2% | 53.5 | 2025-08-07 |
| 16 |  Claude Opus 4.5 | 40.0% | 57.9 | 2025-11-24 |
| 17 | GPT-4o | 40.0% | 43.3 | 2024-11-20 |
| 18 | Claude Sonnet 4.5 | 39.8% | 53.7 | 2025-09-29 |
| 19 | GPT-5 minihigh | 39.0% | 54.6 | 2025-08-07 |
| 20 | Claude 3.7 Sonnet | 39.0% | 49.9 | 2025-02-24 |
| 21 | Gemini 2.5 Flash | 38.0% | 52.7 | 2025-04-17 |
| 22 | Claude Opus 4 | 38.0% | 52.6 | 2025-05-22 |
| 23 | GPT-5 nanohigh | 37.2% | 47.4 | 2025-08-07 |
| 24 | o1medium | 37.0% | 48.1 | 2024-12-17 |
| 25 | GPT-5 nanomedium | 35.4% | 45.5 | 2025-08-07 |
| 26 | Claude Opus 4.1 | 35.0% | 51.4 | 2025-08-05 |
| 27 | Claude Sonnet 4 | 34.0% | 50.5 | 2025-05-22 |
| 28 | GPT-4o mini | 34.0% | 35.4 | 2024-07-18 |
| 29 | Claude 3.5 Sonnet | 33.0% | 47.1 | 2024-10-22 |
| 30 | Gemini 2.5 Flash Lite 09 | 30.0% | 46.9 | 2025-09-25 |
More multimodal benchmarks · All LLM benchmarks
Related benchmarks
Top head-to-heads on VPCT
Cite as: BenchLeader, “VPCT leaderboard”, data as of 6 Nov 2026, https://www.benchleader.com/benchmarks/vpct.
VPCT: questions
- What does VPCT measure?
- Images of physical setups; the model predicts the outcome. Scores are reported in percent of tasks solved; higher is better.
- What is the top score on VPCT?
- As of 6 Nov 2026, the highest score on VPCT is 91.0% by Gemini 3 Pro (result dated 18 Nov 2025), ahead of GPT-5.2 at 84.0%. Scores come from Epoch AI Benchmarking Hub and are re-read daily.
- Which AI model leads VPCT?
- Gemini 3 Pro leads VPCT with 91.0% as of 6 Nov 2026 (result dated 18 Nov 2025), ahead of GPT-5.2 at 84.0%.
- How many models have VPCT results?
- 30 model configurations have a VPCT result on BenchLeader, all taken from Epoch AI Benchmarking Hub.
- Who runs VPCT and how often is it updated?
- VPCT is published by Epoch AI Benchmarking Hub. BenchLeader re-reads the published results once a day and records the date each result was published.
- Does VPCT count toward the BenchLeader Index?
- No. VPCT is shown for reference but left out of the composite index.