Leni Tops Four AI Benchmarks Ahead of OpenAI, Anthropic, Google, and Perplexity

Leni, an AI analytics platform for commercial real estate, achieved leading results across four major AI benchmarks, announced in a press release. The company outperformed systems from OpenAI, Anthropic, Google, and Perplexity AI in tests measuring research quality, reasoning, and reliability.

Leni placed first on the DRACO Benchmark for deep research, developed by Perplexity AI and Harvard, scoring 71.6 percent. It ranked in the top two globally on SpreadsheetBench Verified, completing 365 of 400 spreadsheet tasks correctly. On BullshitBench Version 2, which tests whether AI identifies nonsensical questions, Leni detected 98 percent of fabricated premises, surpassing all 142 public models on the leaderboard.

The platform also led the GAIA benchmark, developed by Meta and Hugging Face, which measures the ability to complete multi step real world tasks. Leni scored 77.0 percent on the validation set, ahead of Genspark, Manus, and OpenAI Deep Research, ranking first across all GAIA difficulty levels.

Leni's system is built around its Universal Data Model, a standardized data framework for multifamily real estate. It integrates data from spreadsheets, PDFs, and core systems to automate investment and asset management workflows. The company says its architecture enables secure, model agnostic automation for commercial real estate teams managing more than $40 billion in assets.