Delivers throughout enterprise, analytical, and scientific work
Enterprise work lives in spreadsheets, PDFs, charts, audio, and video. Neon can synthesize throughout all of those inputs in a single 1 million-token context window and turns inputs into structured deliverables. Impartial evaluations rank Neon as one of many high fashions and on benchmarks just like the Vals Index, it ranks #2 (70.9%), and reaches #1 on each the Authorized Analysis Bench and Tax Agent Bench.
Neon additionally reveals sturdy wins throughout classes on Surge AI, rating #1 on most benchmarks throughout finance, accounting, spreadsheet modeling, and complicated constraint following. It additionally scores 73.5% on Chartography for studying dense skilled charts and performs on par with main fashions on Brokers’ Final Examination v1.1 (75.8%). It achieves #1 total on each Zapier’s AutomationBench (56.3%) and Andon Labs’ Merchandising-Bench 2 ($16,843), a long-horizon simulation of working a enterprise. Lastly, it debuts #2 globally with a rating of 1336 on Enviornment’s new Agent Enviornment Index throughout 4,200+ real-world agent classes.
On scientific and mathematical reasoning, spanning quantitative evaluation, arithmetic, and technical analysis issues, Neon is greatest at school on Terminal-Bench Science (70.5%), RiemannBench (84.0%), and HLE-Diamond (71.3%).







