Claude Opus 4.7 Is Out: What UK Businesses Need to Know

Anthropic pushed out a new flagship model last week, and if your team uses Claude for any kind of serious work, the changes are worth a proper look.

Claude Opus 4.7 went live on 16/04/2026. It is available through Claude.ai, the Claude API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry, with pricing held steady at $5 per million input tokens and $25 per million output tokens (roughly £4 and £20 at current exchange rates).

What actually changed Three areas stand out: coding, vision, and control. On coding, Anthropic reports a 13 per cent lift on standard benchmarks and three times as many production tasks completed successfully compared with Opus 4.6. For any team using Claude Code, or any workflow where Claude writes, reviews, or refactors code, this is a meaningful step forward rather than a marginal one. The model handles long-running, multi-step tasks more reliably, which matters for automated pipelines and large codebases.

On vision, the maximum image resolution has climbed from 1,568 pixels on the long edge to 2,576 pixels, roughly 3.75 megapixels. That translates to better analysis of engineering drawings, product screenshots, detailed charts, and scanned documents. If your team feeds images into Claude to extract data or review designs, the quality of those responses has improved noticeably.

On control, there is a new effort level called xhigh, sitting between high and max. It gives developers finer tuning over the trade-off between reasoning depth and response speed. A Task Budgets feature is also in public beta, designed to help teams keep token spend predictable on extended agent jobs.

The /ultrareview command The most immediately useful addition for developers lives inside Claude Code: a new /ultrareview slash command. Running it on a piece of code opens a focused review session built to catch subtle design flaws and logic gaps that a standard pass tends to miss. Anthropic describes it as simulating a senior reviewer looking at the whole picture rather than scanning line by line. For smaller teams without deep engineering bench depth, that is a practical way to raise quality before a pull request reaches production. It is not a substitute for human review, but it adds a useful extra check at no additional cost beyond normal usage.

Why it matters for UK businesses UK businesses are shifting from exploring AI to depending on it for real work. When a flagship model improves this sharply in a single release, the gap between teams running an up-to-date setup and those still on older defaults grows wider. Finance and legal teams doing document review will benefit from the vision improvements, particularly where source material includes scanned or image-heavy files. Development teams will feel the coding gains most directly, especially those running Claude through Claude Code on a daily basis. Operations teams building automated workflows will find Task Budgets useful for controlling costs as agent jobs grow more complex.

The security changes are worth noting too. Opus 4.7 introduces a Cyber Verification Programme that opens higher-capability security use to verified professionals while blocking the same for unverified requests. For UK businesses in regulated industries, structured access controls like this are a welcome addition. One thing to do this week

If you use Claude through the API, update your model string to claude-opus-4-7. If you are on Claude.ai, the upgrade is automatic. Inside Claude Code, run /ultrareview on something that matters before the end of the week and see what it flags. If you are not yet tracking token spend on longer agent runs, take a look at the Task Budgets beta before your costs start to scale.

At Adevious AI, we build on Claude and Claude Code every day, so updates like this feed directly into the tools and workflows we design for clients. If you want to understand how Opus 4.7 fits into what your business is already doing with AI, or if you are only just starting that conversation, we would be glad to help. Get in touch at adevious.co.uk.