Parable models are trained on real agent work: multi-step tool use, planning, and thinking traces captured from Claude Fable 5 and GPT-5.5 sessions, not synthetic Q&A. Every release is eval-gated against its base before it ships.

This page hosts the IBM Granite 4.1 line of the family. It adds a structured <think> reasoning stage Granite does not ship with, tuned for local coding and terminal work on an enterprise-friendly Apache-2.0 base.

Highlights:

• Trained on real agent traces: multi-step tool use, planning, and <think> reasoning from actual Claude Fable 5 and GPT-5.5 sessions, not synthetic Q&A.
• Eval-gated releases: every model must beat its base on a held-out test split before it ships; strictly graded qualitative results are published, which is rare in this niche.
• Licence: Apache-2.0 weights, ready for local and commercial use (see data provenance note below).

Evaluation (held-out test split, identical eval code for base and fine-tune):

• 8b test loss: 2.030 (base) to 0.617, a 70% reduction
• 3b test loss: 2.824 (base) to 0.376, an 87% reduction (chat-tuned: strongest at explanations and one-liners; prefer the 8b for multi-part scripts)

Strictly graded qualitative results (20 of 34 fully correct, 32 of 34 correct or partial across 34 coding and terminal prompts) are published on the Hugging Face cards; publishing strict numbers is rare in this niche, judge accordingly.

Usage notes

• Responses open with a <think>...</think> block before the final answer.
• Budget at least 2500 tokens for generation; reasoning models think before they answer.
• Sampling: temperature 0.7, top_p 0.95.

Family

• Qwen3 line: parable/qwen3-fable
• Full-precision weights, quant tables, and complete eval reports: Hugging Face, AnkitAI

Get Parable

• Ollama: this page, plus the sibling line and the family flagship parable/fable
• Hugging Face: full weights, GGUF quants, eval reports, datasets
• LM Studio: search “parable”, or paste any Parable Hugging Face URL

Base model: IBM Granite 4.1 (Apache-2.0). Training data: Glint-Research/Fable-5-traces (AGPL-3.0), gpt5.5-terminal (MIT). Third-party assistant traces; providers’ terms may apply to downstream training.