frontier-infra/jebadiah-9b-v2

Jebadiah 9B v2: A System One Decision Model

Jebadiah 9B v2, developed by Frontier Infra, is a 9 billion parameter model built upon the Qwen/Qwen3.5-9B base. Unlike traditional LLMs that generate free-form text, Jebadiah is a "System One" style decision model, meaning it answers typed questions by providing a probability distribution over predefined option labels. This design makes it highly efficient for structured decision-making, returning a single forward pass per question.

Key Capabilities

• Typed Question Answering: Handles three distinct question types:
  • Choice: Selects one option from N possibilities.
  • Noul: Provides a probability for a yes/no statement (P(yes)).
  • Score: Places a state on an ordered rubric.
• Probabilistic Outputs: Delivers decisions with associated probabilities and confidence scores, allowing for nuanced interpretation.
• API Compatibility: Serves a TypeSafe-compatible /v1/systemone endpoint, enabling integration with existing Jev clients.
• Standard Transformers Model: Can be run on various platforms, including CPU, CUDA GPUs, and Apple Silicon (with MLX builds).
• Efficient Inference: Designed for single-hop judgments, providing fast, direct answers without text generation.

Good For

• Automated Decision Systems: Ideal for applications requiring rapid, structured decisions based on input data.
• Classification Tasks: Excels in scenarios where inputs need to be categorized into predefined labels with confidence scores.
• Probabilistic Reasoning: Useful for systems that need to quantify uncertainty in their decisions.
• Resource-Constrained Environments: Its efficient, non-generative nature makes it suitable for deployment where computational resources or latency are critical.
• Integration with Existing Jev Workflows: Seamlessly integrates with systems using the /v1/systemone API standard.