There's a table here showing some "Overall" and "Median" score, but no context on what exactly was tested. It appears to be in the ballpark as the latest models

There's a table here showing some "Overall" and "Median" score, but no context on what exactly was tested. It appears to be in the ballpark as the latest models, but with some cost advantages with the downside of being just as slow as the original r1 (likely lots of thinking tokens). https://www.reddit.com/media?url=https%3A%2F%2Fpreview.redd.it%2Fdeepseek-r1-0528-v0-09patvqurk3f1.jpeg%3Fwidth%3D1080%26format%3Dpjpg%26auto%3Dwebp%26s%3D38580df0e04cb8e29921a6a55b24a7f76b5b30ff https://www.reddit.com/media?url=https%3A%2F%2Fpreview.redd....