Gemma-4-31B-Fable-5-Agent-Distill

Model overview

Gemma-4-31B-Fable-5-Agent-Distill is published by TeichAI as a gemma4 image-text-to-text model. The captured configuration identifies Gemma4ForConditionalGeneration with model type gemma4, and Safetensors metadata reports 32,682,372,656 parameters. According to the model card, it is a Claude Fable 5 distill trained on claude-code traces.

Recorded capabilities

Use cases in the source record

• Coding-agent work, including code generation, debugging, and architecture planning, using the publisher-recommended claude-code or pi harnesses.
• Transformers-based experimentation with the model card's AutoProcessor loading pattern, sampling settings, and thinking-mode controls.

Limitations and unknowns

• No evaluation results were extracted from this record.
• No context-window value was extracted.
• Provider state is historical snapshot data and should be refreshed before being presented as current.
• Training, capability, and benchmark claims are publisher model-card statements and were not independently verified by Ethen.

Source and provenance

Source: TeichAI/Gemma-4-31B-Fable-5-Agent-Distill

Captured: Unknown. Processed: 2026-09-07T19:36:07.174195+00:00.

Gemma 4 31B it - Claude Fable 5 Distilled The following model was trained on claude-code traces, with some chat data provided by the community. I recommend using the model with claude-code or pi, though other harnesses should work without issues. The data for this model was easily extracted, formatted, and masked for training with Teich 📋 Stage Details & Benchmarks Reasoning was left un-touched Benchmarks coming soon Deep Dive Analysis: For more comprehensive insights regarding the base capabilities of the Gemma 4 architecture, please refer to this Analysis Document . 🌟 Core Skills & Capabilities Thanks to its robust base model an…