AI that reads the small print. And checks its own work.
The latest Claude model sees documents in much higher detail and works longer on complex tasks. For businesses, that means fewer errors on paperwork-heavy jobs.
By the Xuru Technologies team · 17 April 2026
Key takeaways
- Claude Opus 4.7 was released on 16 April 2026 across Claude apps, the API and major clouds.
- Maximum image resolution rose to about 3.75 megapixels, roughly three times earlier Claude models.
- Anthropic deliberately reduced its cyber capabilities compared with Mythos Preview and added automatic blocking of high-risk security requests.
Two big jumps matter for everyday business: reading scanned documents and screenshots accurately, and sticking with long tasks without drifting.
What happened
NYU Shanghai's RITS summarised the release, including gains on coding evaluations and fewer document reasoning errors. Google Cloud's Vertex AI documentation listed it for enterprise customers.
What it means for businesses
- Better document handling. Invoices, forms and contracts read more reliably.
- Longer projects. Agents can complete multi-step work with self-checks.
- Safety by default. Built-in limits reduce misuse risks.
Where higher accuracy pays off
| Task | Benefit |
|---|---|
| Invoice and receipt extraction | Fewer manual corrections. |
| Website audits from screenshots | Spots layout and accessibility issues. |
| Contract summaries | Faster review, with a person signing off. |
| Website code updates | Cleaner changes, reviewed before release. |
Frequently asked questions
When was Claude Opus 4.7 released?On 16 April 2026, in Claude apps, the Anthropic API and major cloud platforms.
On 16 April 2026, in Claude apps, the Anthropic API and major cloud platforms.
What is new in Claude Opus 4.7?Stronger coding, about three times higher image resolution, better document reasoning and tuning for long-running tasks.
Stronger coding, about three times higher image resolution, better document reasoning and tuning for long-running tasks.
Should my business switch AI models often?Build workflows so the model can be swapped easily, then test new versions on your own tasks before switching.
Build workflows so the model can be swapped easily, then test new versions on your own tasks before switching.
Put better AI on your paperwork.
Document automation, website AI assistants and workflows built to use the best model for each job.
Websites, payments, digital marketing, AI automation, consultancy and practical technology training — for businesses and aspiring professionals worldwide.
Saturday – Sunday: closed
Leave a Reply