Benchmark Almanac

DAYJOB: Tax

Claude Sonnet 5.5 10.8% leads since October 2026
Not in model cards

Agents take tax assignments from client request to finished return, workpaper or advice, using messy records. A pass needs 95% of the expert rubric criteria.

50 environments with 1,789 files across eight areas of tax practice. The board reports pass@1.

Made by
Surge AI
Released
October 2026
Kind
Finance
Level
Expert
Scored
Graded by a model
Answers
Agent task
Tasks
All public
Strongest model tested
GPT-6 Astra #3 on the capability index
Last model added
October 2026
Scores refreshed
November 12th, 2026
16
Entries on its own leaderboard, 15 models we track

Claude Sonnet 5.5 leads with 10.8%

Best score per entry from the benchmark's own leaderboard, read on November 12th, 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders.

16 entries. Scroll sideways for the rest.

  1. 10.8 Claude Sonnet 5.5
  2. 9.60 Claude Fable 5.1
  3. 9.20 Claude Opus 5.5
  4. 6.00 Muse Wonder 1.0
  5. 5.00 Grok 4.7
  6. 4.80 GPT-6 Astra
  7. 4.00 GPT-6.1 Sol
  8. 3.60 Muse Spark 1.3
  9. 3.20 Gemini 4 Argon
  10. 1.60 Tencent Hy4 preview
  11. 1.20 DeepSeek V4.1 Flash 1017
  12. 0.80 Gemini 3.8 Flash
  13. 0.40 GLM-5.3
  14. 0.0 Kimi K3
  15. 0.0 Qwen 3.8 Max
  16. 0.0 Mistral Large 4
  1. 1 Claude Sonnet 5.5 adaptive/max 10.8%
  2. 2 Claude Fable 5.1 adaptive/max 9.60%
  3. 3 Claude Opus 5.5 adaptive/max 9.20%
  4. 4 Muse Wonder 1.0 max 6.00%
  5. 5 Grok 4.7 xhigh 5.00%
  6. 6 GPT-6 Astra max 4.80%
  7. 7 GPT-6.1 Sol max 4.00%
  8. 8 Muse Spark 1.3 max 3.60%
  9. 9 Gemini 4 Argon high 3.20%
  10. 10 Tencent Hy4 preview high 1.60%
  11. 11 DeepSeek V4.1 Flash 1017 max 1.20%
  12. 12 Gemini 3.8 Flash high 0.80%
  13. 13 GLM-5.3 max 0.40%
  14. 14 Kimi K3 max 0%
  15. 15 Qwen 3.8 Max xhigh 0%
  16. 16 Mistral Large 4 high 0%

Source: Official DAYJOB: Tax leaderboard

#EntryScoreSettingDateSource
1 Claude Sonnet 5.5 10.8% adaptive/max Sep 2026 Official leaderboard
2 Claude Fable 5.1 9.60% adaptive/max Sep 2026 Official leaderboard
3 Claude Opus 5.5 9.20% adaptive/max Sep 2026 Official leaderboard
4 Muse Wonder 1.0 6.00% max Official leaderboard
5 Grok 4.7 5.00% xhigh Sep 2026 Official leaderboard
6 GPT-6 Astra 4.80% max Sep 2026 Official leaderboard
7 GPT-6.1 Sol 4.00% max Sep 2026 Official leaderboard
8 Muse Spark 1.3 3.60% max Sep 2026 Official leaderboard
9 Gemini 4 Argon 3.20% high Sep 2026 Official leaderboard
10 Tencent Hy4 preview 1.60% high Aug 2026 Official leaderboard
11 DeepSeek V4.1 Flash 1017 1.20% max Oct 2026 Official leaderboard
12 Gemini 3.8 Flash 0.80% high Sep 2026 Official leaderboard
13 GLM-5.3 0.40% max Aug 2026 Official leaderboard
14 Kimi K3 0% max Jul 2026 Official leaderboard
15 Qwen 3.8 Max 0% xhigh Aug 2026 Official leaderboard
16 Mistral Large 4 0% high Oct 2026 Official leaderboard

Number one over time

Lab claims count from the day the card came out, independent runs from the model’s release.

Anthropic
Own leaderboard 1 leader
Claude Sonnet 5.5 10.8%

Source: Official DAYJOB: Tax leaderboard

Scores over time

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Best so far
0% 20% 40% 60% 80% 100% Near the ceiling (85%) Claude Sonnet 5.5 Tencent Hy4 preview Kimi K3 0% 20% 40% 60% 80% 100% Near the ceiling (85%) Claude Sonnet 5.5 Tencent Hy4 preview

Source: Official DAYJOB: Tax leaderboard

No lab model card reports it yet

We read the headline results of 522 launch posts from 52 labs, and none of them list this benchmark.

More in Law, finance and medicine

All 179 in Law, finance and medicine
  • MedXpertQA-MM In use Models answer expert-level medical exam questions that combine images with clinical records and test results. Scoring is accuracy. Qwen 3.8 Max 80.4%
  • HealthBench Professional In use 525 physician-authored tasks from real clinician chats (care consult, documentation, research) with physician-adjudicated rubrics. Claude Sonnet 5.5 69.2%
  • HealthBench In use 5,000 realistic health conversations graded against physician-written rubrics. A length-adjusted variant penalizes verbosity. GPT-5 67.2%
  • FinSearchComp In use Agents search the web to fetch live financial data, look up history, and run multi-step investigations, across 635 expert-written questions. Scored by answer accuracy. Ling-3.1-flash 81.7%
  • KBL In use Models answer Korean legal knowledge and reasoning tasks and Korean bar exam questions. It has 7 knowledge tasks, 4 reasoning tasks and 2,510 bar exam items. Solar Pro 4 77.5%
  • KorMedMCQA In use Models answer 7,469 multiple-choice questions from Korean licensing exams for doctors, nurses, pharmacists and dentists, held from 2012 to 2024. Solar Pro 4 93.2%