DAYJOB: Tax — Benchmarking AI Agents on Professional Tax Work
Summary
- Surge AI has introduced DAYJOB: Tax, a new benchmark designed to evaluate AI agents on professional tax preparation, compliance, and advisory services.
- The benchmark includes 50 environments across eight areas of tax practice, featuring 1,789 workspace files such as PDFs, Excel workbooks, and receipts.
- Claude Sonnet 5.5 leads the leaderboard with a 10.8% pass rate at a 95% rubric threshold, followed by Claude Fable 5.1 and Claude Opus 5.5.
- DAYJOB: Tax tests whether AI agents can exercise professional judgment, uncover unstated consequential issues, and handle messy real-world documentation.
Why it matters
Current AI tax tools often focus on isolated calculation or retrieval tasks, whereas DAYJOB: Tax measures end-to-end professional ownership, revealing significant limitations in frontier models regarding unstated risk identification and judgment.