TaxCalcBench
PulseAugur coverage of TaxCalcBench — every cluster mentioning TaxCalcBench across labs, papers, and developer communities, ranked by signal.
-
Law&Order framework achieves 100% accuracy in tax law autoformalization
Researchers have developed a neuro-symbolic framework called Law&Order to automatically convert tax law into executable symbolic programs. This method combines large language model synthesis with cell-level verification…
-
Open Source Tax Engine Achieves Record Score on TaxCalcBench
An open-source tax engine has achieved a record score of 96% on the TaxCalcBench benchmark, outperforming models like GPT Sol and An Ape and a Fox. The engine, which utilizes Claude Sonnet 5, demonstrated superior perfo…
-
Frontier LLMs fail tax calculations; experts advise deterministic engines
A new benchmark, TaxCalcBench, reveals that even frontier Large Language Models struggle with tax calculations, with the best performer, Gemini 2.5 Pro, only getting 32% of tax returns correct. The study suggests that L…