Claude Opus 4.7 topping AI performance charts - link
| German pumo | 04/16/26 | | Pet Relief Area | 04/16/26 | | Guy Debord | 04/16/26 | | ,.,....,..,.,.,,,,..,..,.,..,.,.,.,... | 04/16/26 | | rats pressing the lever | 04/16/26 | | German pumo | 04/16/26 | | Diamond Dallas Chad | 04/16/26 |
Poast new message in this thread
Date: April 16th, 2026 12:08 PM Author: Pet Relief Area
If this translates into real performance increase then that's astonishing.
Sonnet 4.6 is already extremely good
(http://www.autoadmit.com/thread.php?thread_id=5857574&forum_id=2...id#49821342) |
Date: April 16th, 2026 2:46 PM
Author: ,.,....,..,.,.,,,,..,..,.,..,.,.,.,...
Not as good as Mythos but still a substantial improvement over Opus 4.6. The SWE benchmark scores are still rapidly improving. Software engineering is totally fucked soon
(http://www.autoadmit.com/thread.php?thread_id=5857574&forum_id=2...id#49821610) |
Date: April 16th, 2026 4:01 PM Author: Diamond Dallas Chad
Oops...it's actually only better until you have a long cinversation or put a lot of files in context at which point it shits the bed:
https://x.com/DavidOndrej1/status/2044811093005025831
Tokens are also more expensive and it uses more of these more expensive tokens, so actually, it sucks.
(http://www.autoadmit.com/thread.php?thread_id=5857574&forum_id=2...id#49821755) |
|
|