Claude Opus 4.7

Claude Opus 4.7 is a large language model from Anthropic, sold as part of the Opus tier of the Claude family. It is used mainly for coding, agentic workflows and complex knowledge work. The provided pages do not describe Opus 4.7 itself. Anthropic's homepage now promotes newer models (Opus 5.5, Sonnet 5.5, Fable 5.2 and Mythos 5.2), so Opus 4.7 appears to be an earlier generation.

Details here come from creator commentary, not vendor documentation. In Matthew Berman's coverage it scores below GPT-5.5 on DeepSWE, at about three times the cost per trial (around $16 and 37 minutes per trial, with higher token use). It reportedly tends to forget parallel behaviors in multi-part prompts but explores git history well. It was the top model on CursorBench but also the most expensive, at about $11 per task and $30 per million output tokens. Perplexity Computer used it to handle parts of a knowledge-base build while GPT-5.5 did the planning.

Access is through Anthropic's API and Claude products, and likely through third-party platforms and coding tools such as Cursor. Specifications such as context window, exact release date and official pricing were not in the provided material, so they are left unspecified.