DAYJOB: Tax: leaderboard
DAYJOB: Tax (Surge AI, 2026): 50 professional tax assignments across eight areas of practice, from individual returns to partnerships, estates and international tax, with 1,789 workspace files. An agent works on local files in a Linux sandbox through the OpenHands harness and delivers returns, workpapers, reviews or client advice. Unlike TaxBench or TaxCalcBench, which pose a defined question or return, "DAYJOB also asks whether the client asked the right question": the agent must find issues the client overlooked. Score: pass@1 at a 95% rubric threshold, the share (%) of attempts that satisfy at least 95% of the task's rubric criteria (five attempts per task, Claude Opus 4.8 autograder); reasoning effort as labelled. Higher is better. Source: https://surgehq.ai/benchmarks/dayjob-tax
Metric: Pass@1 at a 95% rubric threshold (%). Source: surgehq.ai. Saturation forecast: Around 2030. 16 models tracked.
Top models
| # | Model | Score | Overall rank |
|---|---|---|---|
| 1 | Claude Fable 5.1 (Max) | 9.6 | #10 (Claude Fable 5.1) |
| 2 | Muse Wonder (Max) | 6 | #18 (Muse Wonder) |
| 3 | Grok 4.7 (xHigh) | 5 | #26 (Grok 4.7) |
| 4 | GPT-6 Astra (Max) | 4.8 | #6 (GPT-6 Astra) |
| 5 | GPT-6.1 Sol (Max) | 4 | #7 (GPT-6.1 Sol) |
| 6 | Muse Spark 1.3 (Max) | 3.6 | #31 (Muse Spark 1.3) |
| 7 | Gemini 4 Argon (High) | 3.2 | #9 (Gemini 4 Argon) |
| 8 | Hy4 preview (High) | 1.6 | #45 (Hy4 preview) |
| 9 | DeepSeek V4.1 Flash (1017) (Max) | 1.2 | #27 (DeepSeek V4.1 Flash (1017)) |
| 10 | Gemini 3.8 Flash (High) | 0.8 | #30 (Gemini 3.8 Flash) |
| 11 | GLM-5.3 (Max) | 0.4 | #44 (GLM-5.3) |
| 12 | Kimi K3 (Max) | 0 | #35 (Kimi K3) |
| 13 | Mistral Large 4 (High) | 0 | #106 (Mistral Large 4) |
| 14 | Qwen 3.8 Max (xHigh) | 0 | #34 (Qwen 3.8 Max) |
Interactive version: theaggregate.ai/benchmark?slug=dayjob-tax · How It Works · Data refreshed daily, snapshot 2026-11-05.