Opus 4.7 Improves Coding

Anthropic has released Claude Opus 4.7 to general availability. The key point is that it performs better than Opus 4.6 on difficult software engineering tasks, follows instructions more reliably even on long tasks, and delivers higher output quality.

Pricing remains unchanged. It stays at $5 per million input tokens and $25 per million output tokens, and is available across the Claude product lineup and major partner clouds. Anthropic described Opus 4.7 as less general-purpose than Mythos Preview but stronger than Opus 4.6.

• Vision: processes higher-resolution images
• Output: improved quality for UI layouts, documents, and slides
• Instruction following: focused on reducing drift and constraint deviation on long-running tasks
• Security: includes guardrails that block high-risk cyber requests

In practice, teams should run long refactors, multi-file changes, and failure-recovery tasks against their internal codebases to verify whether supervision costs actually decrease.