Anthropic's latest disclosure shows that Claude now generates over 80% of production merge code for the company, a significant increase from single-digit percentages before its release in February 2025. This indicates that AI is becoming more deeply integrated into actual software development workflows.
The code share has increased to over 80%.
The company disclosed that Claude has now become the primary code generation tool in its internal engineering processes. Here, "production code" refers to code that has been merged and deployed to the live environment, not draft content in testing stages.
- Currently, production code accounts for more than 80%.
- Before the release, the percentage was only in the single digits.
- The comparison time point is around February 2025.
Can continuously handle long-duration tasks
Anthropic also stated that Claude can now complete software engineering tasks lasting up to 12 hours continuously. These tasks often involve more than just code completion—they may include debugging, modifying, testing, and multi-step collaboration.
This means the use cases for the model are expanding from short-term Q&A and snippet-based programming to more comprehensive engineering workflows.
The company discusses the future evolution of AI.
In this update, Anthropic also outlines several possible development scenarios, including a slowdown in AI progress and recursive self-improvement of AI, where AI systems assist in advancing the development of the next generation of AI.
Supporters argue that such enhanced capabilities could accelerate progress in fields like medicine, mathematics, and scientific research. However, critics point out that this更像是编程辅助工具的持续增强，还不足以证明通用人工智能已经接近实现。
Additional information: The data mentioned in this article comes from Anthropic’s own disclosures and primarily reflects usage within its internal development processes, not independent third-party measurements.