Anthropic’s Claude Opus 4.7 is here: Check improvements, safety features, and what’s new

Anthropic has released Claude Opus 4.7, the new powerful artificial intelligence (AI) model that succeeds Opus 4.6. This model is designed for everyday tasks such as writing, coding assistance, research, and workflow automation, with improved performance in coding, reasoning, and safety-focused capabilities. 

The new AI model comes days after Anthropic launched Claude Mythos, its powerful cybersecurity model, released in preview for limited users. Whereas the new Claude Opus 4.7 models serve general users who rely on more accessible and balanced alternatives focused on usability and reliability. 

What is Claude Opus 4.7? 

Claude Opus 4.7 is Anthropic’s new all-purpose AI model, which is built to support handling a wide range of everyday and professional tasks. According to Anthropic’s blog post, the new AI model brings upgrades to coding, multidisciplinary reasoning, and overall reliability. The company said the “Opus 4.7 handles complex, long-running tasks with rigor and consistency, pays precise attention to instructions, and devises ways to verify its own outputs before reporting back.”

The Claude Opus 4.7 also comes with cyber capabilities and built-in safeguards that can detect risky or harmful requests. The company also claims to automatically block misuse related to cybersecurity. It is also using Opus 4.7 to test how well its safety systems work in the real world. 

Security experts also use Opus 4.7 for cybersecurity purposes such as vulnerability research, penetration testing, and red-teaming. Professionals can join Anthropic’s Cyber Verification Program to test the model.

Claude Opus 4.7 performance

Anthropic claims that Claude Opus 4.7 delivers impressive performance benchmarks in coding, reasoning, and tool use. The model scored 64.3% on agentic coding (SWE-bench Pro), up from 53.4% in Opus 4.6, and achieved 87.6% on the verified version of the same benchmark.

Opus 4.7 demonstrated improved tool usage and computer interaction capabilities, scoring 77.3% in scaled tool use and 78% in agentic computer use. However, in areas like agentic search, it trails behind competitors such as GPT-5.4. 

In cybersecurity capabilities, the model remains slightly below Opus 4.6, but significantly behind Mythos Preview. 

The company further claims that Opus 4.7 delivers improved instruction-following, supports high-resolution image generation, performs better in finance agent evaluations, and enhances its ability to utilise file system-based memory.

Claude Opus 4.7 availability 

The Claude Opus 4.7  is rolling out across all Claude products, including API, Amazon Bedrock, Google Cloud’s Vertex AI, and Microsoft Foundry.