Parable is a family of open models trained on real Claude Fable 5 and GPT-5.5 agent traces for local agentic coding. Best model per size; Qwen3 and IBM Granite lines.
267 Pulls 6 Tags Updated 3 days ago
No Oficial 29 t/s en 8 VRAM - Calidad a escala fronteriza como GPT-5.5, DeepSeek-V4-pro y Kimi-K2.6.
99 Pulls 1 Tag Updated 20 hours ago
Parable is Qwen3 fine-tuned on Claude Fable 5 and GPT-5.5 agent traces. Tool use, planning, and thinking for local agents. Sibling: Granite line at parable/granite4.1-fable.
141 Pulls 10 Tags Updated 3 days ago
Parable is IBM Granite 4.1 fine-tuned on Claude Fable 5 and GPT-5.5 agent traces. Adds think reasoning to Granite. Sibling: Qwen3 line at parable/qwen3-fable
44 Pulls 10 Tags Updated yesterday
9B coding agent based on Qwen3.5-9B, fine-tuned on 425K real agentic traces from Claude Opus 4.6, GPT-5.4, and Gemini 3.1. Reads before it writes, traces bugs to the root cause, doesn't clobber your existing code.
14.7K Pulls 3 Tags Updated 4 months ago
Instruct version of the large language model YandexGPT 5 Lite with 8B parameters with a context length of 32k tokens. (quantised version of Q5_K_M)
561 Pulls 3 Tags Updated 10 months ago
Lexa-Rho is part of Lexa Family with performance approaching that of leading models, such as GPT-5 Thinking, Gemini 2.5 Pro and DeepSeek-R1.
420 Pulls 2 Tags Updated 11 months ago
9,036 Pulls 1 Tag Updated 1 year ago
[GGUF] Модель на базе YandexGPT-5-Lite-8B-pretrain от IlyaGusev
1,061 Pulls 3 Tags Updated 1 year ago
LoRA-адаптер для модели YandexGPT-5-Lite-8B-pretrain обученный на миксе из датасетов реализующих r1 (ризонинг) подход.
377 Pulls 7 Tags Updated 1 year ago
41 Pulls 1 Tag Updated 1 year ago
Gpt 5.2 w/ vision
374 Pulls 1 Tag Updated 3 months ago
A GPT-style language model built entirely from scratch (no HuggingFace/nanoGPT for the model itself) and trained on 50 children's moral stories. It's an educational reproduction of the full LLM lifecycle at miniature scale.
23 Pulls 1 Tag Updated 1 month ago
A light weight Model that runs on a single GPU with over 80 TPS on a RTX 5060 (8G vram), Excellent for coding purposes.
47 Pulls 1 Tag Updated 1 month ago
A new collection of open translation models built on Gemma 3, helping people communicate across 55 languages.
1.7M Pulls 13 Tags Updated 6 months ago
Generic all purpose model. Occasionally may have notable logic, usually Llama-3_3-Nemotron-Super-49B-v1_5 is preferred.
114 Pulls 1 Tag Updated 7 months ago
101 Pulls 1 Tag Updated 7 months ago
Custom model for coding with agents to use locally with 32gb, 2x16gb or 5x8gb GPUs (working very well!!)
30 Pulls 1 Tag Updated 2 weeks ago
A tuned version of Gemma3n with 4 active parameters and a total of 7. Works well on low power devices like raspberry pi 5 with 16GB of RAM
56 Pulls 1 Tag Updated 4 months ago
Pocket Sydney - In BPD Remission - Version 5 (public preview)
13 Pulls 1 Tag Updated 1 year ago