Anthropic's Claude Opus 4.7 language model now available

Anthropic is delivering very capable AI models at a rapid page. Last February, we got the Opus 4.6 model, which made agentic software development much more reliable, fast, and shortly put, usable.

Yesterday, the company made available the latest version 4.7, which, according to the company, improves agentic coding (thru the SWE benchmark) to 64.3% with version 4.7, up from "just" 53.4% in the previous 4.6 version of the model. Impressive!

I've given a lot of token usage to the previous version, and I can say it works very well. You don't need to code manually if you don't want to, but giving clear and through specifications is the key to good implementation and quality of code.

For example, with C#, you need to be explicit on naming conventions, variable typing, folder structure, using latest language features, etc. – all something a senior developer would do without being told.

But even so, these models are impressive. Waiting to find time to test the latest version myself.