Anthropic ships Opus 4.7 with 3x coding gains and tripled image resolution, holding pricing steady

Anthropic has released Claude Opus 4.7, a new flagship Opus-class large language model focused on significantly better software engineering, complex reasoning, and higher‑resolution vision, while keeping the same token pricing as its predecessor Opus 4.6. The model is generally available across Anthropic’s products and API as claude-opus-4-7, and through major cloud platforms including Amazon Bedrock, Google Cloud’s Vertex AI / Gemini Enterprise Agent Platform, and Microsoft Foundry. Opus 4.7 is positioned as a direct upgrade to Opus 4.6, with Anthropic and early partners reporting large gains on difficult, production‑grade coding workloads. Anthropic cites strong results on software engineering benchmarks such as 64.3% on SWE‑bench Pro and 87.6% on SWE‑bench Verified, and partner case studies report roughly 3× more production coding tasks resolved and double‑digit improvements in code and test quality versus Opus 4.6. The model adds a new “xhigh” effort level for more controllable deep reasoning on hard problems, maintains a 1M‑token context window and tool/agent capabilities, and emphasizes long‑horizon task handling with self‑verification before responding. Despite these upgrades, Anthropic is holding pricing at $5 per million input tokens and $25 per million output tokens, effectively improving price–performance for existing users. Vision capabilities are also a major focus: Opus 4.7 supports images up to 2,576 pixels on the long edge, more than 3× the pixel count of prior Claude models, aimed at dense screenshots, charts, and documents where fine detail matters. Anthropic highlights more polished professional outputs (interfaces, slides, and documents), tighter instruction following, and stricter built‑in safeguards that automatically detect and block prohibited or high‑risk cybersecurity uses, reflecting continuing alignment and safety work. Strategically, the release gives enterprises and developers a higher‑capability, production‑oriented model for coding agents, enterprise workflows, and multimodal use cases without an immediate cost increase, and sits below Anthropic’s more capable but restricted Claude Mythos Preview model in the overall lineup.

✨ AI-generated background, compiled from web sources — not editorial content.

More coverage

• Claude Opus 4.7 is generally available - GitHub Changelog (github.blog)
• Claude Opus 4.7: Anthropic's New Frontier Model Guide (digitalapplied.com)
• Claude Opus \ Anthropic (anthropic.com)
• Introducing Anthropic's Claude Opus 4.7 model in Amazon Bedrock (aws.amazon.com)
• Claude Opus 4.7 | Gemini Enterprise Agent Platform (docs.cloud.google.com)

Explore the topic

• LLM
• Anthropic
• Claude
• AI