Anthropic launches Claude Opus 4.7 with improvements in coding and vision
Anthropic has announced Claude Opus 4.7, describing it as an improvement over Opus 4.6 in software engineering, vision, and longer complex tasks.
Summary
Anthropic announced on April 16, 2026 the release of Claude Opus 4.7. The company presents the new model as a notable improvement over Opus 4.6, especially in advanced software engineering and the hardest tasks.
According to Anthropic, Opus 4.7 handles complex long-running work with greater rigor, follows instructions more precisely, and finds ways to verify its own outputs before responding. The company also says the model has better vision, allowing it to interpret higher-resolution images.
In practice
Anthropic highlights gains in coding, multi-step work, and professional tasks that require more consistency. The company says users are able to hand off harder technical work with more confidence than before.
The model also accepts images up to 2,576 pixels on the long edge, increasing the visual detail available for tasks such as interface analysis, document reading, chart interpretation, and other complex visual work.
What we still don't know
While Anthropic presents Opus 4.7 as an upgrade over Opus 4.6, it also points to two practical considerations: an updated tokenizer that can change token counts for the same input and higher output usage at higher effort levels.
The company also says the safety profile is similar to Opus 4.6, with improvements on some measures and modest weaknesses on others.
Why it matters
- It reinforces competition among frontier models for coding, agents, and longer technical work.
- It shows that differentiation is increasingly tied to reliability and consistency in real workflows.
- It expands multimodal usefulness with support for higher-resolution images.
- It may be especially relevant for teams using AI in demanding technical workflows.