Claude Opus 4.7: Anthropic’s new productivity model

Anthropic has unveiled its new Claude Opus 4.7 model. Although it is not as powerful as Mythos, it outperforms the other available models in an assessment focused on real-world tasks.

Competition between AI labs is intense, forcing them to release new models at a rapid pace. This week, Anthropic is launching Claude Opus 4.7. While this release is not as capable as Claude Mythos in cybersecurity, it represents a major step forward for people who use AI to automate office work or write computer code. Earlier Claude models already have an excellent reputation among businesses, and Anthropic is adding new strengths to its tools with Opus 4.7.

A new productivity tool

For coding, Claude Opus 4.7 can independently carry out tasks that previously required human oversight. For other office duties, the new AI benefits from improved vision, allowing it to “see” user-submitted documents at a higher resolution. Opus 4.7 is also better at following instructions, making it easier for users to get the results they want. Anthropic says this new AI follows directions literally, meaning that some prompts that were ineffective with earlier models may deliver good results with this one.

For people using AI to create documents or visuals, Anthropic also says Claude Opus 4.7 is more creative, improving the quality of its output. Claude Opus 4.7 could also excel in finance, as it surpasses every other AI currently available on a benchmark designed specifically for that sector. Most importantly, Claude Opus 4.7 is the new leader on GDPval-AA, ahead of OpenAI and Google models. This assessment measures an AI’s ability to handle everyday “real-world” tasks, spanning 44 occupations across 9 business sectors. In other words, this new AI positions itself as the best tool for automating repetitive work.

A trial before the release of Claude Mythos

This April, Anthropic also introduced an ultra-powerful AI model called Claude Mythos. However, unlike Opus 4.7, which is already available through the start-up’s products, Claude Mythos is currently limited to a selected group of organisations. The reason? Mythos already has highly advanced capabilities for finding security vulnerabilities and could therefore lead to a wave of cyberattacks.

Nevertheless, Anthropic intends to make Mythos available to the public when it can do so safely. The launch of Claude Opus 4.7 gives the company an opportunity to test measures it may apply to Claude Mythos to stop hackers using the AI to carry out cyberattacks. According to Anthropic, the public version of Claude Opus 4.7 includes “safeguards that automatically detect and block requests that indicate prohibited or high-risk cybersecurity uses”. “What we learn from the real-world deployment of these safeguards will help us achieve our ultimate goal: broad deployment of Mythos-class models”, it adds.

Businesses will, however, be able to obtain an unrestricted version of Claude Opus 4.7 for legitimate purposes. They will need to enrol in a special programme.

What we think

Anthropic has made businesses and professionals its niche. It is therefore unsurprising that, in developing this new model, the company concentrated on the qualities that matter most to its target audience. This is supported by an independent benchmark that seeks to assess the productivity gains delivered by AI.

Everything suggests that Anthropic’s main rivals, Google and OpenAI, could follow a similar path. While having an AI that is popular with the public is a success, businesses and professionals working with a chatbot are more likely to pay for a subscription.