| Legal agents need current, authoritative sources to answer accurately. The evidence that matters is scattered across court dockets, state codes, agency guidance, and government publications, often on sites that general web search reaches poorly. | Parallel for Legal upgrades our index and ranking for legal agents across all Parallel APIs, so agents can find hard-to-reach primary sources across jurisdictions and return citable evidence. It's live today, with no new plan or subscription. | The cost-accuracy frontier for legal | | PWS-Legal: On 238 private legal questions, Parallel leads on accuracy and holds the entire cost-accuracy frontier, beating every other provider at up to 2.3x lower cost. LegalBench Citation: Parallel leads every other search API on accuracy, at the same or lower cost. Legal Hallucinations: Parallel matches OpenAI and Exa's accuracy at 2.6x lower cost.
| Public benchmarks and their answers end up online and in training data. To account for "benchmaxxing" bias, we built PWS-Legal: 238 held-out questions with verified answers, including 60 about jurisdictions outside the US and 33 in 12 languages other than English. | Learn more about PWS-Legal. | Powering legal AI at scale | Harvey uses Parallel to reach legal sources across 60+ countries. GC AI brings authoritative legal sources into cited answers for more than 2,200 in-house legal teams. Finch cut per-query costs by 90% after moving to the Task API. | Want to see it work? Our open-source cookbook turns a CSV of states and one legal question into a cited, side-by-side jurisdiction survey. | For the full results and methodology, read the blog. | | |
|