Fastino Labs Releases Finance and Healthcare Models Built on NVIDIA Nemotron

Fastino Labs announced in a press release two specialized open weight models for finance and healthcare, both post trained on NVIDIA Nemotron 3.5 Lightning. Fastino-Nemotron-3.5-Lightning-Finance and Fastino-Nemotron-3.5-Lightning-Healthcare are available on Hugging Face under the Apache 2.0 license.

The company said the finance model raises FinQA accuracy from 15.86% for the base model to 59.23%, with gains across financial reasoning benchmarks. The healthcare model showed gains on MEDEC and MedCalc-Bench, and improved HealthBench performance from 40.62% to 48.17% across 700 reserved clinical conversations.

Both models were trained by the Fastino Fine-Tuning Agent, which the company is releasing in private preview. Fastino said the agent takes a plain language task description and handles research, data curation, evaluation set creation, parallel training runs, recovery from failed runs, data contamination tests, and checkpoint selection.

The finance model is built for numerical reasoning, research, and summarization over financial documents such as filings, reports, and tables. The healthcare model is built for clinical conversation quality, medical reasoning, note summarization, discharge documentation, and medical concept extraction from unstructured text.