# Tax Agent Bench leaderboard

Agentic research on US tax questions, about 60% federal corporate income tax. The rest covers partnership, state and local, transfer pricing and employment tax. Agents use tools and must answer with citations.

## Tax Agent Bench: accuracy % (all-pass x citation factor) (higher is better, 10 models)

| Rank | Model | Maker | Score | Metric | Setting | Reported by | Source | URL | Date |
| --- | --- | --- | --- | --- | --- | --- | --- | --- | --- |
| 1 | Qwen 3.8 Max 0902 | Qwen | 49.7 | accuracy % (all-pass x citation factor) | test set | Official board | Vals.ai leaderboard | https://www.vals.ai/benchmarks/tax_agent_bench | 29 Oct 2026 |
| 2 | Claude Fable 5.1 | Anthropic | 49.2 | accuracy % (all-pass x citation factor) | test set | Official board | Vals.ai leaderboard | https://www.vals.ai/benchmarks/tax_agent_bench | 29 Oct 2026 |
| 3 | Gemini 4 Carbon | Google | 48.9 | accuracy % (all-pass x citation factor) | test set | Official board | Vals.ai leaderboard | https://www.vals.ai/benchmarks/tax_agent_bench | 29 Oct 2026 |
| 4 | Claude Opus 5.5 | Anthropic | 45.6 | accuracy % (all-pass x citation factor) | test set | Official board | Vals.ai leaderboard | https://www.vals.ai/benchmarks/tax_agent_bench | 29 Oct 2026 |
| 5 | Claude Opus 5 | Anthropic | 45.5 | accuracy % (all-pass x citation factor) | test set | Official board | Vals.ai leaderboard | https://www.vals.ai/benchmarks/tax_agent_bench | 29 Oct 2026 |
| 6 | Gemini 4 Argon | Google | 44.6 | accuracy % (all-pass x citation factor) | test set | Official board | Vals.ai leaderboard | https://www.vals.ai/benchmarks/tax_agent_bench | 29 Oct 2026 |
| 7 | DeepSeek V4.1 Flash 1017 | DeepSeek | 44.4 | accuracy % (all-pass x citation factor) | test set | Official board | Vals.ai leaderboard | https://www.vals.ai/benchmarks/tax_agent_bench | 29 Oct 2026 |
| 8 | Wonder Full | Not stated | 43 | accuracy % (all-pass x citation factor) | test set | Official board | Vals.ai leaderboard | https://www.vals.ai/benchmarks/tax_agent_bench | 29 Oct 2026 |
| 9 | Muse Spark 1.3 Max | Meta | 42.4 | accuracy % (all-pass x citation factor) | test set | Official board | Vals.ai leaderboard | https://www.vals.ai/benchmarks/tax_agent_bench | 29 Oct 2026 |
| 10 | Claude Sonnet 5.5 | Anthropic | 42.0 | accuracy % (all-pass x citation factor) | test set | Official board | Vals.ai leaderboard | https://www.vals.ai/benchmarks/tax_agent_bench | 29 Oct 2026 |

Scores are never combined across benchmarks. Each row names its metric, source and date.

## About this benchmark

- Maker: Vals AI
- Year: Not stated
- Use case: Finance knowledge
- Size: 193 test questions
- Public split: A public split exists (it includes example P-004), but its size is not stated
- Grading: All-pass on expert rubric checks, scaled by citation quality: score = all-pass x (0.7 + 0.3 x citation quality). Weighted Pass is the secondary metric. An LLM judge grades.
- Licence: proprietary
- Submission: Not stated
- Products allowed: Not stated
- Latest version: 2026-10-29
- What it does not cover: US only. No personal tax filing and no non-US jurisdictions.
- Last checked: 1 Nov 2026
- Links: site https://www.vals.ai/benchmarks/tax_agent_bench, leaderboard https://www.vals.ai/benchmarks/tax_agent_bench
- Directory page: https://benchmarks.finance/benchmarks/vals-tax-agent-bench/
- Leaderboard: https://benchmarks.finance/leaderboards/vals-tax-agent-bench/

Source: https://benchmarks.finance/leaderboards/vals-tax-agent-bench/

Created and maintained by Moneda.