| domain | vals.ai |
| summary | This website showcases top-performing models based on the Vals Index, a multimodal evaluation tool. The site presents three standout models:
1. Claude Sonnet 4.5 Thinking - Vals Index Score: 66.7 2. GPT 5 - Vals Index Score: 62.3 3. Claude Haiku 4.5 Thinking - Vals Index Score: 59.9
These models are ranked using the Vals Multimodal Index, an alternative to traditional benchmarks like accuracy. The site also introduces GLM 4.6, another model evaluated on all benchmarks, with a score of 4.6.
The website emphasizes the importance of better model evaluation methods, as current benchmarks are deemed insufficient. Users can view more models and join a mailing list for benchmark updates by contacting the site's creators. |
| title | Home |
| description | Astro description |
| keywords | index, thinking, benchmarks, models, haiku, open, sonnet, model, coding, score, cost, accuracy, performing, tasks, finance, weight, performance |
| upstreams |
|
| downstreams |
|
| nslookup | A 76.76.21.21 |
| created | 2025-11-10 |
| updated | 2025-11-10 |
| summarized | None |
|
|