# granite-4.1-3b-llmon-preview

> Model Summary: Granite-4.1-3B-LLMON-preview is a 3B parameter instruct model fine-tuned from Granite-4.1-3B-base to consume prompts in LLMON (LLM Object Notation) instead of a conventional chat template.

- **Organisation:** ibm-granite
- **Category:** Chat & Text Generation
- **Parameters:** 3.7B
- **Context length:** 131,072 tokens
- **Licence:** Apache 2.0 (Commercial use OK)
- **Released:** 2026-10-01
- **Hugging Face:** 2.2K downloads (30 days), 18 likes

## Estimated memory

| Quant | ≈ GB |
|---|---|
| Q2_K | 2.3 |
| Q3_K_M | 2.87 |
| Q4_K_M | 3.29 |
| Q5_K_M | 3.67 |
| Q6_K | 4.07 |
| Q8_0 | 4.93 |
| F16 | 8.28 |

## Runs with

- [llama.cpp](https://selfhost.directory/project/llamacpp)
- [SGLang](https://selfhost.directory/project/sglang)
- [Transformers](https://github.com/huggingface/transformers)
- [vLLM](https://selfhost.directory/project/vllm)

## About

Model Summary: Granite-4.1-3B-LLMON-preview is a 3B parameter instruct model fine-tuned from Granite-4.1-3B-base to consume prompts in LLMON (LLM Object Notation) instead of a conventional chat template. LLMON wraps every part of a prompt (system policy, instructions, data, documents, tool definitions, and tool results) in an individually addressable block and adds an exec block that names which instruction to run, which inputs it operates on, and where the output goes.

• Developers: Granite Team, IBM
• HF Collection: Granite Experiments HF Collection
• Other sizes: Granite-4.1-8B-LLMON-preview
• Technical Blog: LLMON: An LLM-Native Markup Language, now in Granite 4.1 3B and 8B
• GitHub Repository: generative-computing/llmon
• Visualization Demo: LLMON on Hugging Face Spaces
• Release Date: November 12, 2026
• License: Apache 2.0

Supported Languages: English, German, Spanish, French, Japanese, Portuguese, Arabic, Czech, Italian, Korean, Dutch, and Chinese. Users may fine-tune the model for languages beyond these.

Intended use: The model is designed to follow general instructions and can serve as the foundation for AI assistants across diverse domains, including business applications, as well as for LLM agents equipped with tool-use capabilities. It is particularly suited to settings where a prompt mixes trusted instructions with untrusted material (retrieved documents, pasted content, or tool results) and the boundary between the two must be explicit.

## Alternatives

- [nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16](https://selfhost.directory/models/nvidia/nvidia-nemotron-3-nano-4b-bf16) (4B, nvidia-nemotron-open-model-license, ≈4.17 GB at Q4): NVIDIA-Nemotron-3-Nano-4B-BF16 is a small language model (SLM) trained from scratch by NVIDIA, and designed as a unified model for both reasoning and non-reasoning tasks. It responds to user queries and tasks by first generating a reasoning trace and then concluding with a final response.
- [ibm-granite/granite-4.1-3b](https://selfhost.directory/models/ibm-granite/granite-4.1-3b) (3.4B, Apache 2.0, ≈3.14 GB at Q4): Model Summary: Granite-4.1-3B is a 3B parameter long-context instruct model finetuned from Granite-4.1-3B-Base using a combination of open source instruction datasets with permissive license and internally collected synthetic datasets.
- [Qwen/Qwen3-4B-Instruct-2507](https://selfhost.directory/models/qwen/qwen3-4b-instruct-2507) (4B, Apache 2.0, ≈4.01 GB at Q4): We introduce the updated version of the Qwen3-4B non-thinking mode, named Qwen3-4B-Instruct-2507, featuring the following key enhancements:
- [openbmb/MiniCPM5-2B](https://selfhost.directory/models/openbmb/minicpm5-2b) (2.5B, Apache 2.0, ≈2.32 GB at Q4): We are releasing MiniCPM5-2B, the second model in the MiniCPM5 series, following MiniCPM5-1B. It is a dense 2B Transformer that scales up the same training recipe, built for on-device, local deployment, and resource-constrained scenarios, reaching 2B-class open-source SOTA.
- [ibm-granite/granite-4.2-3b](https://selfhost.directory/models/ibm-granite/granite-4.2-3b) (3.7B, Apache 2.0, ≈3.29 GB at Q4): Granite is a family of open-source large language models developed by IBM, designed for enterprise and research use. Granite models are built to be versatile, safe, and efficient — covering a range of sizes and capabilities from compact edge-deployable models to large-scale reasoning systems.
- [Qwen/Qwen3-4B](https://selfhost.directory/models/qwen/qwen3-4b) (4B, Apache 2.0, ≈4.01 GB at Q4): Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models.

More: https://selfhost.directory/models/ibm-granite/granite-4.1-3b-llmon-preview/alternatives

## Runs well on

- [GeForce RTX 4060](https://selfhost.directory/models/ibm-granite/granite-4.1-3b-llmon-preview/on/rtx-4060): Q8_0, est. 38 tok/s
- [Apple M4 (16 GB)](https://selfhost.directory/models/ibm-granite/granite-4.1-3b-llmon-preview/on/apple-m4-16gb): F16, est. 9 tok/s
- [Radeon RX 6700 XT](https://selfhost.directory/models/ibm-granite/granite-4.1-3b-llmon-preview/on/rx-6700-xt): F16, est. 29 tok/s
- [GeForce RTX 3060 12GB](https://selfhost.directory/models/ibm-granite/granite-4.1-3b-llmon-preview/on/rtx-3060): F16, est. 27 tok/s
- [Arc B580](https://selfhost.directory/models/ibm-granite/granite-4.1-3b-llmon-preview/on/arc-b580): F16, est. 34 tok/s
- [Tesla T4](https://selfhost.directory/models/ibm-granite/granite-4.1-3b-llmon-preview/on/tesla-t4): F16, est. 24 tok/s

## Similar size

- [meta-llama/Llama-3.2-3B-Instruct](https://selfhost.directory/models/meta-llama/llama-3.2-3b-instruct) (3.2B)
- [Qwen/Qwen3-4B](https://selfhost.directory/models/qwen/qwen3-4b) (4B)
- [Qwen/Qwen2.5-3B-Instruct](https://selfhost.directory/models/qwen/qwen2.5-3b-instruct) (3.1B)
- [microsoft/Phi-4-mini-instruct](https://selfhost.directory/models/microsoft/phi-4-mini-instruct) (3.8B)
- [google/gemma-2-2b-it](https://selfhost.directory/models/google/gemma-2-2b-it) (2.6B)
- [Qwen/Qwen3-4B-Instruct-2507](https://selfhost.directory/models/qwen/qwen3-4b-instruct-2507) (4B)

## FAQ

**How much VRAM does granite-4.1-3b-llmon-preview need?**
About 3.3 GB at Q4_K_M with an 8K context. Q8_0 needs about 4.9 GB and full precision about 8.3 GB. Longer contexts need more memory for the KV cache.

**Can granite-4.1-3b-llmon-preview run on the RTX 4060?**
Yes. It fits at Q8_0 (about 4.9 GB), at an estimated 38 tokens/s. It fits comfortably on the Apple M4 (16 GB), Radeon RX 6700 XT, RTX 3060 12GB.

**Can I use granite-4.1-3b-llmon-preview commercially?**
Yes. It is released under Apache 2.0. Permissive: use, modify and sell freely; keep the notice and licence text.

**What is the context length of granite-4.1-3b-llmon-preview?**
131,072 tokens (128K). The full window needs far more memory than an 8K context.

**How do I run granite-4.1-3b-llmon-preview with llama.cpp?**
Run llama-server -hf liodon-ai/granite-4.1-3b-llmon-preview-GGUF:Q4_K_M. It downloads the GGUF and starts an OpenAI-compatible server.

## Links

- huggingface: https://huggingface.co/ibm-granite/granite-4.1-3b-llmon-preview
- paper: https://arxiv.org/abs/2610.38120
- demo: https://huggingface.co/spaces/ibm-granite/LLMON-Attention-Visualization

---
Source: https://selfhost.directory/models/ibm-granite/granite-4.1-3b-llmon-preview