Anthropic ships Claude Opus 4.7
Anthropic released Claude Opus 4.7 with reported improvements for software‑engineering tasks, higher‑resolution vision, and better instruction following while keeping pricing unchanged. The update was announced on social channels as a step forward for coding and multimodal workflows. (x.com)
Anthropic released Claude Opus 4.7 on April 16, saying the new model is generally available across its apps, API, and major cloud partners. (anthropic.com) Anthropic said Opus 4.7 improves on Opus 4.6 in advanced software engineering, especially on harder coding tasks that previously needed closer human supervision. The company also said the model follows instructions more precisely and can check its own work before answering. (anthropic.com) The update also raises image resolution for vision tasks, which Anthropic said helps the model read denser screenshots and other visual inputs. Pricing stays at $5 per million input tokens and $25 per million output tokens, the same rates Anthropic listed for Opus 4.6-era pricing. (anthropic.com 1) (anthropic.com 2) Large language models are systems trained to predict the next piece of text, and companies now tune them for coding, document work, and image understanding as well as chat. Anthropic has been pushing Claude toward long-running “agent” tasks, where the model can work through multi-step jobs over extended sessions. (anthropic.com) Anthropic’s Claude 4 launch in May 2025 framed Opus as its top coding model and added features such as tool use, file access, and stronger memory for agent workflows. Opus 4.7 extends that line rather than creating a new product tier. (anthropic.com) (support.claude.com) The release also fits Anthropic’s recent split between broadly available products and more restricted frontier systems. Anthropic said Opus 4.7 is “less broadly capable” than Claude Mythos Preview, which it has kept limited while testing cybersecurity controls under Project Glasswing. (anthropic.com) (cnbc.com) For Opus 4.7, Anthropic said it added safeguards that automatically detect and block prohibited or high-risk cybersecurity requests. The company also opened a Cyber Verification Program for security researchers doing work such as vulnerability research, penetration testing, and red-teaming. (anthropic.com) Opus 4.7 is available in Claude products, through the Claude API, and on Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry. Anthropic’s release notes list the launch on April 16, 2026, as the latest model update. (anthropic.com) (support.claude.com)
