Patsnap launches first AI patent novelty search benchmarking tool

*This article is adapted from * Asia IP

Singapore-based global innovation intelligence company Patsnap launched PatentBench, a quantitative benchmarking tool for AI patent novelty search, on September 10, 2025.

The first tool of its kind in the industry, Patsnap PatentBench enables consistent and qualitative evaluation and comparison of AI tools used by IP professionals and research and development units.

“PatentBench, developed by Patsnap, is the industry’s first most comprehensive benchmark testing mechanism for artificial intelligence in IP. It goes beyond testing the AI models for text-based generation but drills down to deep, technical and legal jargon often used in patent work,” revealed Patsnap founder and CEO Jeffrey Tiong.

With this capability, the patent benchmark evaluates the performance of three AI tools: Patsnap’s Novelty Search AI Agent, ChatGPT-o3 (with web search) and DeepSeek-R1 (with web search).

As of its launch, the benchmark may be used for novelty searches.

“Built on an open-source base model, the agent has been systematically refined with specialized patent knowledge, allowing it to understand the nuances of patent language and search logic. By integrating Retrieval Augmented Generation (RAG), it combines real-time data retrieval with generative capabilities, enabling high-quality, low-hallucination search results. This empowers the agent to accurately identify key technical features, apply precise search strategies and outperform general-purpose models in professional novelty search tasks,” Tiong explained.

“In our conversations with customers and prospects, many have shared the need for a patent benchmark across AI tools but struggled to find one accurate enough that caters specifically to IP,” he added.

According to him, general large language model (LLM) benchmarks are not robust enough to assist in patent work. The task requires deep expertise across three domains: law, technology and business. “Oftentimes, they miss the nuances that professionals care about in claims, are unfamiliar with the diverse and evolving regulations across patent offices and lack evaluation standards tailored to real-world needs like patent searches,” said Tiong.

Eventually, PatentBench will cover freedom-to-operate search, patent invalidity, invention disclosure drafting and patent translation benchmarks.

For the latest news on Vertex Ventures SE Asia and India and our portfolio companies, follow us on Linkedin or subscribe to our monthly newsletter.