multimodal benchmark · included in ECI
VPCT
VPCT benchmark (score column: Correct).
Results
34
Random baseline
33.3%
Score ceiling
100%
Released
30 Jan 2025
Score by model release date
Best score by organisation
Leaderboard
| # | Model | Score | Relative | Setting | Released | Run | Evidence |
|---|---|---|---|---|---|---|---|
| 1 | Gemini 3 Pro Preview Google DeepMind | 91.0% | — | 18 Nov 2025 | — | Source ↗ | |
| 2 | GPT-5.2 (xhigh) OpenAI | 84.0% | xhigh | 11 Dec 2025 | — | Source ↗ | |
| 3 | Gemini 3 Flash Preview Google | 72.6% | — | 17 Dec 2025 | — | Source ↗ | |
| 4 | GPT-5.2 OpenAI | 67.0% | high | 11 Dec 2025 | — | Source ↗ | |
| 5 | GPT-5 OpenAI | 66.0% | high | 7 Aug 2025 | — | Source ↗ | |
| 6 | GPT-5 (medium) OpenAI | 63.2% | medium | 7 Aug 2025 | — | Source ↗ | |
| 7 | GPT-5.1 OpenAI | 58.7% | high | 13 Nov 2025 | — | Source ↗ | |
| 8 | o4-mini (medium) OpenAI | 57.5% | medium | 16 Apr 2025 | — | Source ↗ | |
| 9 | GPT-5.1 (medium) OpenAI | 53.3% | medium | 13 Nov 2025 | — | Source ↗ | |
| 10 | o3 (medium) OpenAI | 52.0% | medium | 16 Apr 2025 | — | Source ↗ | |
| 11 | Gemini 2.5 Pro Preview (Mar 2025) Google DeepMind | 48.0% | — | 31 Mar 2025 | — | Source ↗ | |
| 12 | Gemini 2.5 Pro Preview (Jun 2025) Google DeepMind | 46.4% | — | 5 Jun 2025 | — | Source ↗ | |
| 13 | Gemini 2.5 Flash (Sep 2025) Google DeepMind | 46.2% | — | 25 Sep 2025 | — | Source ↗ | |
| 14 | GPT-4.5 Preview (Feb 2025) OpenAI | 45.0% | — | 27 Feb 2025 | — | Source ↗ | |
| 15 | gemini-robotics-er-1.5-preview | 40.8% | — | 26 Sep 2025 | — | Source ↗ | |
| 16 | GPT-5 mini (medium) OpenAI | 40.2% | medium | 7 Aug 2025 | — | Source ↗ | |
| 17 | Claude Opus 4.5 (32k thinking) Anthropic | 40.0% | 32K | 24 Nov 2025 | — | Source ↗ | |
| 18 | GPT-4o (Nov 2024) OpenAI | 40.0% | — | 20 Nov 2024 | — | Source ↗ | |
| 19 | Claude Sonnet 4.5 (32k thinking) Anthropic | 39.8% | 32K | 29 Sep 2025 | — | Source ↗ | |
| 20 | GPT-5 Mini OpenAI | 39.0% | high | 7 Aug 2025 | — | Source ↗ | |
| 21 | Claude 3.7 Sonnet Anthropic | 39.0% | — | 24 Feb 2025 | — | Source ↗ | |
| 22 | Claude Sonnet 4.5 (no thinking) Anthropic | 38.0% | — | 29 Sep 2025 | — | Source ↗ | |
| 23 | Claude Opus 4 Anthropic | 38.0% | 16K | 22 May 2025 | — | Source ↗ | |
| 24 | Gemini 2.5 Flash Preview (Apr 2025) Google DeepMind | 38.0% | — | 17 Apr 2025 | — | Source ↗ | |
| 25 | GPT-5 Nano OpenAI | 37.2% | high | 7 Aug 2025 | — | Source ↗ | |
| 26 | o1 (medium) OpenAI | 37.0% | medium | 17 Dec 2024 | — | Source ↗ | |
| 27 | GPT-5 nano (medium) OpenAI | 35.4% | medium | 7 Aug 2025 | — | Source ↗ | |
| 28 | Claude Opus 4.1 Anthropic | 35.0% | — | 5 Aug 2025 | — | Source ↗ | |
| 29 | Claude 3.7 Sonnet (64k thinking) Anthropic | 35.0% | 64K | 24 Feb 2025 | — | Source ↗ | |
| 30 | Claude Sonnet 4 Anthropic | 34.0% | 32K | 22 May 2025 | — | Source ↗ | |
| 31 | GPT-4o-mini OpenAI | 34.0% | — | 18 Jul 2024 | — | Source ↗ | |
| 32 | Claude 3.5 Sonnet (Oct 2024) Anthropic | 33.0% | — | 22 Oct 2024 | — | Source ↗ | |
| 33 | Claude 3.5 Sonnet (Jun 2024) Anthropic | 33.0% | — | 20 Jun 2024 | — | Source ↗ | |
| 34 | Gemini 2.5 Flash-Lite (Sep 2025) Google DeepMind | 30.0% | — | 25 Sep 2025 | — | Source ↗ |
Caveats: settings (reasoning effort, agent scaffold) differ between rows and materially affect scores; the best reported setting per model is shown. Source: Epoch AI, CC BY 4.0.