Introducing Claude Opus 4.5

1. 0:00 – 0:06

   Sholto introduces Claude Opus 4.5 as Anthropic’s top model

   1. SPSpeaker0:00

      Hey, I'm Sholto. I'm an engineer and researcher here at Anthropic. Claude Opus 4.5 is our best model yet. It's

2. 0:06 – 0:12

   Reliability and intuition: “it just gets it”

   1. SPSpeaker0:06

      the best in the world at coding, agentic tasks, and everyday work like spreadsheets. What's harder to show is how it just gets it.

3. 0:12 – 0:18

   Fewer interventions: longer autonomous runs on tasks

   1. SPSpeaker0:12

      I trust this model more. My time between interventions keeps getting longer, and I keep seeing colleagues say it solved a bug that

4. 0:18 – 0:24

   Efficiency via deliberation: knowing when to think before acting

   1. SPSpeaker0:18

      Sonnet never could have found. The model is also more efficient because it knows when to think before acting so that it makes the right change.

5. 0:24 – 0:30

   Benchmark highlight: top score on a two-hour take-home engineering task

   1. SPSpeaker0:25

      We've got this take-home. It's a two-hour intensive engineering task, and in that time,

6. 0:30 – 0:47

   Improved frontend and vision for better computer use

   1. SPSpeaker0:30

      the model scored higher than any human ever has. It's also much better at front end, it's way better at vision, and that makes it a lot better at using computers. Opus 4.5 is available today. For the first time, it's on every major cloud platform. Let us know what you think. [outro music]

Episode duration: 0:49