Avertry/BibleAI

VISIONPricing:Input $0.4 / Cached $0.08 / Output $0.8Concurrent Unit Cost:1Model Size:7.9BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 4, 2026License:apache-2.0Architecture:Transformer Open Weights Featherless Exclusive Cold

BibleAI is a 7.9 billion parameter Gemma 4 E4B model developed by rhemabible, specifically refined for questions related to the Bible, theology, church history, and faith. It underwent a comprehensive CPT, SFT, and DPO training pipeline to ensure high-integrity, citation-oriented responses. This model excels at providing concise and accurate answers within its specialized domain, avoiding fabrication of verses or facts. Its primary use case is supporting Bible study and theological inquiry with reliable information.

Loading preview...

BibleAI: Specialized Theological Language Model

BibleAI is a 7.9 billion parameter model built on the Gemma4ForConditionalGeneration architecture, developed by rhemabible. It has been meticulously refined through a three-stage training process: CPT (Continued Pre-Training), SFT (Supervised Fine-Tuning), and DPO (Direct Preference Optimization), making it highly specialized for religious and theological queries.

Key Capabilities and Training

  • Domain Specialization: Optimized for questions concerning the Bible, theology, church history, and general faith topics.
  • Training Data: SFT stage utilized 15,289 examples, while DPO involved 967 preference pairs, ensuring alignment with desired response styles.
  • Response Policy: Adheres to a strict policy to provide concise, accurate answers, specifically listing items from verses without commentary, and avoiding fabrication of facts or verses.
  • Integrity: Designed to deliver high-integrity, citation-oriented responses, explicitly stating uncertainty when applicable.

Intended Use Cases

  • Bible Study: Assisting users with scripture-centered theological support.
  • Church History & Faith Q&A: Providing factual answers on historical and faith-related inquiries.
  • High-Integrity Information: Ideal for applications requiring reliable, non-fabricated information within its domain.