[Claude Code Review Tiers: The More Expensive, the Better at Finding Bugs] According to monitoring by Beating, Anthropic has revamped the /code-review process for Claude Code. Users can now choose from Low, Medium, High, X-high, and Ultra tiers. Different tiers employ distinct review processes, no longer relying solely on the same set of prompts to think harder. Low performs a quick scan of code changes. Medium reviews and verifies from multiple angles. High engages independent sub-agents to identify and validate issues. X-high additionally checks the impact of changes on other parts of the code. Official self-testing shows that Low identifies 17% of bugs, compared to 8% for unnamed competitors. X-high improves detection to 25%, with a single review costing nearly $2. The highest tier, Ultra, launches a group of review agents in the cloud to independently reproduce and verify each issue. It typically runs for 5 to 10 minutes. Pro and Max users each receive 3 one-time free credits, after which the service usually costs $5 to $20 per session. [Original Link]