Claude Opus 4.7 Is Out: Better Long-Run Coding, Stronger Vision, Same Price

TL;DR

• Anthropic launched Claude Opus 4.7 with stronger performance on long-running engineering tasks and vision-heavy workflows.
• Availability is broad on day one: Claude API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry.
• Pricing stays at Opus 4.6 levels: $5 per million input tokens and $25 per million output tokens.
• Anthropic also ships additional cybersecurity controls for higher-risk use cases.

Claude Opus 4.7 is positioned as a practical upgrade for teams that need dependable execution across multi-step coding and reasoning. In the official launch note, Anthropic describes improvements in consistency and self-verification, which matters most in production workflows where a model must keep context and quality over longer sessions, not just solve one-shot prompts.

Vision capability is another focus area in this release. As detailed in the Claude Opus 4.7 System Card, the model is described as better at interpreting complex visual inputs and producing cleaner professional outputs, including interface- and slide-like artifacts. For teams building multimodal assistants, this lowers the amount of manual clean-up after generation.

On safety, Anthropic links Opus 4.7 to an explicit cybersecurity risk-control path, including the Cyber Verification Program. The broader risk-reduction direction is also framed alongside Project Glasswing, with the goal of supporting legitimate security research while reducing misuse in high-risk scenarios.

Summary

• Opus 4.7 is a meaningful reliability upgrade for teams doing long, multi-step engineering work with AI.
• Anthropic kept pricing flat versus Opus 4.6, making migration easier for existing Opus users.
• If your workflow needs strong coding plus vision in one model, this release is a high-priority candidate to evaluate.

Source: Introducing Claude Opus 4.7