frontier-infra/jebadiah-27b

VISIONPricing:Input $1.6 / Cached $0.15 / Output $12Concurrent Unit Cost:2Model Size:27BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Sep 26, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

Jebadiah 27B is Frontier Infra's 27 billion parameter open System One style decision model, built on Qwen3.8-27B. It is specifically designed to answer typed questions with a probability distribution over option labels, rather than generating free-form text. This model excels at structured decision-making tasks, offering choice, noul (yes/no), and score question types, making it suitable for applications requiring probabilistic, interpretable outputs.

Loading preview...

Jebadiah 27B: A Specialized Decision Model

Jebadiah 27B, developed by Frontier Infra, is a 27 billion parameter open System One style decision model. Unlike traditional LLMs that generate text, Jebadiah is engineered to provide probabilistic answers to structured questions, making it highly effective for automated decision-making processes. It is built upon the Qwen/Qwen3.8-27B base model.

Key Capabilities

  • Typed Question Answering: Processes questions of three types: choice (select one from N options), noul (yes/no statements with a probability of 'yes'), and score (placing a state on an ordered rubric).
  • Probabilistic Outputs: Returns a probability distribution over option labels for each question, offering interpretable and quantifiable decisions.
  • System One Compatibility: Serves a TypeSafe-compatible /v1/systemone endpoint, allowing integration with existing Jev clients.
  • Strong Performance: Achieved a Decision Index 0.2.1 score of 54.67, ranking No. 5 among 70 open models (as of 2026-09-28 board data).
  • Flexible Deployment: Can be run as a standard transformers model, with support for vLLM, local servers, and various quantized formats like GGUF and MLX for Apple silicon.

Good For

  • Automated Decision Systems: Ideal for applications requiring structured, probabilistic decisions, such as customer support routing, risk assessment, or content moderation.
  • Interpretable AI: Provides clear probability scores for each decision, enhancing transparency and auditability.
  • Resource-Constrained Environments: While the 27B model requires significant memory (approx. 56 GB for bf16), smaller 9B and 4B versions are available for more constrained local deployments.
  • Integration with Existing Jev Clients: Seamlessly integrates with systems using the Jev client by simply changing the endpoint.

Finetunes
10 models