SAN FRANCISCO — Independent developers and AI research groups on Sunday published early assessments of Anthropic's newly released Claude Opus 5, testing the company's claims that the model excels at complex, multi-step coding and agentic tasks.

Anthropic announced Claude Opus 5 earlier in the week, describing it as a "thoughtful and proactive" model built to handle extended software engineering workflows. The company positioned the release as its most capable system to date, targeting enterprise customers and developers who use AI agents for automated coding.

Early hands-on reports circulated across developer forums and benchmarking platforms, with users comparing Opus 5's performance on tasks such as SWE-bench and long-horizon reasoning against OpenAI and Google models. Some testers highlighted gains in code reliability, while others questioned pricing and latency for large-scale deployment.

The launch intensifies competition among frontier AI labs. OpenAI and Google DeepMind are expected to respond with model updates in the coming weeks, as Anthropic increasingly focuses on the enterprise coding market, where Claude has gained traction through tools like GitHub integrations and its Claude Code product.

An Anthropic spokesperson said the company would continue publishing safety evaluations and technical documentation for Opus 5, with customer access expanding through its API and cloud partners in the days ahead.