Anthropic's Claude Opus 4.1 aims at GPT-5 with programming boost
Anthropic has unveiled Claude Opus 4.1, an enhanced iteration of its flagship hybrid language model, signaling the company’s intent to remain competitive as OpenAI prepares for the launch of GPT-5. The update, released on August 5, 2025, introduces targeted improvements across key performance areas.
The new model delivers notable advancements in programming, demonstrating superior capabilities in code refactoring, data-intensive analysis, and agentic functions—the ability to independently manage complex, multi-step tasks. Claude Opus 4.1 is now accessible to paying users through the Claude platform, Claude Code, and via its API, as well as on Amazon Bedrock and Google Cloud Vertex AI. Pricing for the updated model remains consistent with the previous Opus 4 version, and developers can access it using the claude-opus-4-1-20250805 API tag.
Claude Opus 4.1 has set a new benchmark on the SWE-bench Verified test, achieving a score of 74.5 percent. This represents an improvement of approximately two points over its predecessor, Opus 4, and roughly five points ahead of OpenAI’s o-series models. The SWE-bench Verified benchmark assesses an AI model’s proficiency in identifying and resolving real-world bugs within open-source codebases. Beyond programming, the model also shows gains in analysis and research tasks, with Anthropic noting enhanced detail tracking and agent-style search capabilities. The company highlights that Claude Opus 4.1 outperforms other leading AI models in areas such as agentic coding, visual reasoning, and math competitions. Furthermore, the now-defunct coding startup Windsurf reported that Claude Opus 4.1 achieved a one-standard-deviation improvement on its internal benchmark for junior developers, a leap comparable to the transition from Sonnet 3.7 to Sonnet 4.
The timing of Claude Opus 4.1’s release is particularly strategic, coinciding with widespread anticipation for OpenAI’s GPT-5. Reports suggest GPT-5 is expected to raise performance standards in programming, mathematics, and agent-based tasks, though it is not anticipated to deliver the monumental leap seen between GPT-3 and GPT-4.
With GPT-5 projected to offer incremental gains, Anthropic’s latest update positions the company to maintain its competitive standing. Anthropic is recommending that all users migrate from Opus 4 to Opus 4.1, while also promising “substantially larger” improvements in the weeks to come, underscoring its commitment to ongoing innovation in the rapidly evolving AI landscape.