New Claude Sonnet 4.5 from Anthropic

We’ve Launched Sonnet 4.5, and by All Measures, It’s Now the Best AI for Developers. But the most exciting part isn’t just its intelligence—it’s its endurance. Many reviewers have particularly highlighted how long it can operate without interruption. At Anthropic, they conducted an internal test where it built a chat application on the level of Slack or Microsoft Teams from scratch. It produced 11,000 lines of code and completed the entire project independently. This took about 30 hours. That’s four times longer than any of their previous AIs could manage. The model didn’t just write code for thirty hours straight. It set up databases, registered domains, and even passed a rigorous SOC 2 audit on its own. Specialists from Canva, who tested it in beta mode, praise it for handling complex and very large-scale tasks—from edits to their core codebase to developing new features and conducting research.

Key Changes in Anthropic Claude Sonnet 4.5

• The biggest improvement is that it has become much better at grasping the essence and context. If you give it a complex task that requires analyzing information and drawing conclusions, it handles it more accurately and doesn’t lose the thread of reasoning.
• What’s especially important for developers is that it has significantly leveled up in code handling. Now it not only writes code more efficiently but also fixes errors more thoughtfully, truly understanding the issue at hand.
• They’ve also done great work with large documents. The model now processes lengthy texts more stably, without “forgetting” what was at the beginning by the time it reaches the end. This is crucial for serious tasks.
• And for developers, there’s a nice bonus: they’ve updated the API platform. Integrating this model into new projects and products is now noticeably simpler and more convenient. Overall, it’s a very solid and practical update.

Update Context

The new model Sonnet 4.5 is available through the Claude Pro subscription, as well as via API for enterprise clients and developers. These changes are aimed at strengthening its position in the commercial AI usage segment. For more details on this tool, read our Claude AI review. Cost of Access to Claude Sonnet 4.5. A key question for users is how much does Claude Sonnet 4.5 cost? The model is available in the Claude Pro subscription, priced at $20 USD per month. For commercial use via API, there’s a pay-as-you-go model, with details available on the official Anthropic website.

Conclusion

Here’s what it all adds up to. With the release of Claude Sonnet 4.5, it’s clear this isn’t just another minor update—it’s a major leap forward for the entire Claude lineup. Its endurance is particularly impressive—when a model can independently manage a complex project for 30 hours straight, from writing code to setting up the full infrastructure. The model’s ability to handle complex projects autonomously for 30 hours, from start to finish. It seems the team at Anthropic has bet on reliability and sustained performance without glitches—and this could become their key advantage. For developers and companies, there’s now an opportunity to entrust AI with truly demanding tasks that previously required constant human oversight and involvement. At the same time, the model remains affordably priced at just $20 per month for the Pro version. Now it’s intriguing to see how competitors will respond, and when similar enhancements will roll out to other models in their lineup, like the lighter Haiku. I think the community is eagerly awaiting that.