Anthropic on 5 August introduced Claude Opus 4.1, a mid-cycle upgrade to its most powerful large language model. The company says the new version delivers sharper performance on agentic tasks, complex reasoning and real-world coding, while keeping the same pricing as Claude Opus 4. In internal testing, Opus 4.1 achieved 74.5% on the SWE-bench Verified software-engineering benchmark, eclipsing both its predecessor and rival models such as OpenAI’s o3 and Google’s Gemini 2.5 Pro. Anthropic also reports higher precision in multi-file code refactoring and a lower refusal rate when responding to user requests. The model is immediately available to paid Claude users through the Claude Code interface, Anthropic’s API, Amazon Bedrock and Google Cloud’s Vertex AI. Developers can access it under the identifier “claude-opus-4-1-20250805.” Anthropic said it expects to release “substantially larger improvements” to its model family in the coming weeks, underscoring the escalating competition in generative AI ahead of OpenAI’s anticipated GPT-5 launch.