Fastino Labs announces two open-weight models for finance and healthcare, post-trained on NVIDIA’s Nemotron 3.5 Lightning

Fastino Labs said it has released two specialized open-weight models aimed at enterprise use cases in finance and healthcare, positioning the move as a way for developers to adapt foundation-model capabilities to regulated, domain-specific tasks. In announcing the models, Fastino Labs said they are “post-trained” on NVIDIA’s Nemotron 3.5 Lightning. Post-training is the stage after a base model is initially trained, where additional tuning is performed to improve performance on targeted tasks, formats, or instruction styles. The company also described its development approach as “entirely using an agent,” implying that the post-training process was guided or executed through an automated agent workflow rather than primarily through manual steps. Fastino Labs did not, in the announcement as reported, provide additional technical details such as the agent’s role boundaries, evaluation criteria, or the specific datasets used for the finance and healthcare specializations. Fastino’s announcement comes as open-weight models continue to draw attention from organizations that want more control over deployment and customization. Open-weight typically means model weights are available for users with the right tooling and permissions, allowing fine-tuning, integration into private environments, or modification for particular data and workflows. For finance-focused deployments, specialized models are often used to assist with tasks such as document understanding, summarization, and structured extraction from reports and internal records. For healthcare, similar capabilities can be applied to clinical documentation workflows, where output formatting, traceability, and careful alignment to domain instructions are particularly important. Fastino Labs is also known for the GLiNER model family, according to the context around its work that has been covered by technology media. The new releases announcement an emphasis on building models that can be tailored to specific vertical needs rather than offering only general-purpose language tools. NVIDIA’s Nemotron 3.5 Lightning is referenced in the announcement as the post-training foundation for both models. The Lightning phrasing is consistent with the company’s broader push to optimize model performance and efficiency, but the Fastino Labs post as reported did not break out which components of Nemotron were used or what performance targets were achieved in the finance versus healthcare variants.

keyFacts":[