Claude Opus 4.7 Release Features (April 2026)
Claude Opus 4.7 launched on April 16, 2026 with features that directly addressed each major Opus 4.6 complaint, including a new X-high effort level, an /ultrareview command, overhauled long-context retrieval, and a 232-page system card.
Anthropic released Claude Opus 4.7 on April 16, 2026. The release notes and system card described features that mapped one-to-one against the most prominent complaints about Claude Opus 4.6. New capabilities: - X-high effort, a 4.7-exclusive effort level above the previous maximum. - /ultrareview, a dedicated review-session command for deeper code review passes. - Long-context retrieval 'completely overhauled,' with higher multi-needle scores deep into documents (relevant to users running the model at 1M-token context). - A new tokenizer, which implies a new base model rather than a fine-tune of 4.6. - A 232-page system card (substantially longer than prior releases). - A new Claude desktop app shipped alongside the model; Theo Browne reported finding more than 40 bugs in the app within an hour. Reported performance gains: - +13% on an internal 93-task coding benchmark. - 3x production task resolution on the Rakuten benchmark. - Visual reasoning rose from 69.1% to 82.1%. Pricing and limits: token cost rose roughly 1x to 1.3x because of the new tokenizer. Boris Cherny stated that Max plan rate limits were increased to offset the per-token cost rise. See Stella Laurenzo's AMD Telemetry Analysis of Claude Opus 4.6 Regression for the regression these features were designed to answer, and Chain-of-Thought Prompting: How Step-by-Step Reasoning Improves LLM Accuracy for background on effort levels and reasoning depth.