
Samaya AI introduces Sterling for financial research
The AMW Read
Sterling adds a specialized financial research model, but its reported benchmark advantage remains vendor-supplied and specific to one evaluation.
Samaya AI introduces Sterling for financial research
Samaya AI has released Sterling, a model designed to plan financial research, call tools, gather evidence, and write reports. Sterling is a post-trained version of Nvidia's Nemotron 3.5 Lightning, a mixture-of-experts model with 30 billion total parameters and 3 billion active per token. It runs within the Samaya platform, where customers must request access or book a demo. The source does not describe a public API or downloadable weights.
The release adds a specialized model to the market for AI tools sold into financial workflows. Samaya reports that Sterling scored 58.9% on its FrontierFinance benchmark, compared with 55.9% for Claude Fable 5.1. It also reports a cost of $0.61 per question for Sterling and $6.89 for Fable 5.1. Those are vendor-reported results on a company-built benchmark, so they do not establish a general performance or cost advantage across financial research tasks.
For finance teams and investors, the model should be evaluated together with its research system. Sterling's setup includes 37 finance-specific tools, retrieval models, and components that shorten tool results and research traces. Buyers should test report quality, evidence handling, and total cost on their own workflows, while checking which tools and settings apply to each comparison. The source says the FrontierFinance dataset is open-source, but does not establish independent validation of the reported scores.