VISTA
Vision-language understanding. Scale AI.
- Published by
- Scale AI SEAL
- Category
- Multimodal
- Index weight
- 1.0
- Models
- 57
- Data as of
- 9 Sept 2026
Scale AI SEAL Leaderboards.
- 1Gemini 2.5 Pro54.6%
- 2GPT-5.4 Pro53.9%
- 3GPT-5 Pro52.4%
- 4o4-mini51.8%
- 5o4-mini51.7%
- 6o3-pro51.6%
- 7Gemini 3 Pro51.5%
- 8GPT-5.450.9%
- 9GPT-5 mini50.4%
- 10o350.1%
- 11GPT-549.7%
- 12o349.6%
- 13Gemini 2.5 Flash49.1%
- 14Claude Sonnet 4.548.8%
- 15Claude Opus 4.148.4%
57 of 57
| # | |||||
|---|---|---|---|---|---|
| 1 |  Gemini 2.5 ProGoogle | 54.6% | preview, march-2025 | 54.1 | – |
| 2 |  GPT-5.4 ProOpenAI | 53.9% | – | 64.2 | – |
| 3 | GPT-5 ProOpenAI | 52.4% | – | 60.7 | – |
| 4 | o4-miniOpenAI | 51.8% | high, april-2025 | 48.0 | – |
| 5 | o4-miniOpenAI | 51.7% | medium, april-2025 | 52.4 | – |
| 6 | o3-proOpenAI | 51.6% | high, june-2025 | 54.4 | – |
| 7 | Gemini 3 ProGoogle | 51.5% | – | 61.6 | – |
| 8 | GPT-5.4OpenAI | 50.9% | – | – | – |
| 9 | GPT-5 miniOpenAI | 50.4% | – | 55.8 | – |
| 10 | o3OpenAI | 50.1% | high, april-2025 | 52.9 | – |
| 11 | GPT-5OpenAI | 49.7% | – | 59.5 | – |
| 12 | o3OpenAI | 49.6% | medium, april-2025 | 53.6 | – |
| 13 | Gemini 2.5 FlashGoogle | 49.1% | preview, may-2025 | 48.8 | – |
| 14 |  Claude Sonnet 4.5Anthropic | 48.8% | – | 52.8 | – |
| 15 | Claude Opus 4.1Anthropic | 48.4% | – | 58.1 | – |
| 16 | Claude 3.7 SonnetAnthropic | 48.2% | thinking, feb-2025 | 52.1 | – |
| 17 | o1-proOpenAI | 47.3% | march-2025 | – | – |
| 18 | Claude Opus 4Anthropic | 47.0% | – | 54.4 | – |
| 19 | Gemini 3.1 Flash LiteGoogle | 46.9% | – | 54.3 | – |
| 20 | GPT-5.2OpenAI | 46.6% | – | 61.5 | – |
| 21 | Claude Opus 4.5Anthropic | 46.4% | – | 60.6 | – |
| 22 | Claude Opus 4.6Anthropic | 46.1% | – | 54.1 | – |
| 23 | Gemini 2.0 FlashGoogle | 45.5% | – | 45.1 | – |
| 24 | Claude Sonnet 4Anthropic | 45.5% | – | 54.0 | – |
| 25 | Claude Opus 4.6Anthropic | 45.5% | – | 55.3 | – |
| 26 | GPT-4.1OpenAI | 45.3% | GPT-4.1 | 47.3 | – |
| 27 | Claude Opus 4.5Anthropic | 45.3% | – | 58.0 | – |
| 28 | o1OpenAI | 45.3% | december-2024 | 53.2 | – |
| 29 | Claude Opus 4.1Anthropic | 45.3% | – | 52.6 | – |
| 30 | Claude Sonnet 4.5Anthropic | 45.0% | – | 52.3 | – |
| 31 | GPT-5.1OpenAI | 43.8% | – | 55.2 | – |
| 32 | Claude Opus 4Anthropic | 43.5% | Claude Opus 4 | 50.7 | – |
| 33 | Gemini 2.0 ProGoogle | 43.3% | preview, feb-2025 | 47.0 | – |
| 34 | Claude Sonnet 4Anthropic | 43.2% | Claude Sonnet 4 | 49.5 | – |
| 35 | Claude 3.7 SonnetAnthropic | 43.0% | february-2025 | 49.8 | – |
| 36 | GPT-4.5OpenAI | 42.1% | preview, february-2025 | 49.6 | – |
| 37 |  Kimi K2.5Moonshot AIopen | 41.9% | kimi-k2.5 | 56.1 | – |
| 38 | GPT-4.1 miniOpenAI | 41.1% | GPT-4.1 mini | 44.5 | – |
| 39 | Gemini 2.0 FlashGoogle | 40.0% | – | 42.3 | – |
| 40 | Claude 3.5 SonnetAnthropic | 38.7% | october-2024 | 45.1 | – |
| 41 |  Llama 4 MaverickMetaopen | 38.3% | Llama 4 Maverick | 42.6 | – |
| 42 | GPT-4oOpenAI | 38.0% | november-2024 | 42.3 | – |
| 43 | Gemini 1.5 ProGoogle | 37.1% | Gemini 1.5 Pro | 42.9 | – |
| 44 | GPT 5.1 InstantOpenAI | 34.9% | – | – | – |
| 45 |  Mistral Medium 3Mistral AI | 34.6% | Mistral Medium 3 | 44.7 | – |
| 46 | Gemini 1.5 Flash 002Google | 34.0% | Gemini 1.5 Flash 002 | 38.7 | – |
| 47 | Pixtral Large (latest)Mistral AIopen | 33.9% | november-2024 | 42.2 | – |
| 48 | Gemini 2.0 Flash-LiteGoogle | 32.7% | – | 47.0 | – |
| 49 |  Qwen2 Vl 72BAlibaba | 28.6% | – | – | – |
| 50 | Claude 3 OpusAnthropic | 27.8% | Claude 3 Opus | 38.9 | – |
| 51 | GPT-4.1 nanoOpenAI | 26.6% | GPT-4.1 nano | 38.1 | – |
| 52 |  Nova ProAmazon | 26.3% | – | 41.0 | – |
| 53 | Pixtral 12BMistral AIopen | 26.0% | september-2024 | – | – |
| 54 | Amazon Nova LiteAmazon | 25.5% | – | 39.4 | – |
| 55 | Llama 3.2 90BMetaopen | 24.6% | Llama 3.2 90B Vision Instruct | 37.5 | – |
| 56 | Llama 3.2 Instruct 11B (Vision)Metaopen | 20.5% | – | 36.4 | – |
| 57 |  Phi 3.5 VisionMicrosoft | 15.2% | – | – | – |