Beta · Model Pilot
Transcription
Beta · Model Pilot
Transcription
General-purpose AI models often apply content filters that can't distinguish legitimate clinical documentation (forensic notes, abuse and trauma assessments) from genuinely harmful material, and block it. Our Lumen models are tuned to reduce these false blocks so your clinical documentation isn't interrupted.
Use of this system is covered under the BastionGPT Business Associate Agreement (BAA) and is secure for use with HIPAA-regulated data such as PHI and PII.
These models are still in active evaluation. They do not yet have published safety or clinical-reliability data and have not received final approval from the BastionGPT Clinical Review team. Review output for accuracy and bias before any clinical use. Please report any errors or concerning output to [email protected].
The Lumen Beta includes three models. Pick one for each chat using the model selector:
During the model pilot, response times can slow down during heavy use and the system may occasionally be unavailable during frequent updates. Additionally, features such as saved prompts and chat history are disabled. Your existing saved prompts and chat history have not been altered and can still be found in your normal BastionGPT account. Final release of the Lumen series of models into your primary BastionGPT account is expected Q4 2026.