Bielik-11B-v3.1-Instruct GGUF

Bielik-11B-v3.1-Instruct GGUF

Chat model that answers in Polish, English and six other languages, packed as GGUF files in five sizes for llama.cpp or Ollama. For people who want an instruct model on their own computer.

Made by speakleash on Hugging Face. These notes are Mindket's, written from its model page and model card; the model's own files and license are on Hugging Face.

About

What it is: This repository holds GGUF copies of Bielik-11B-v3.1-Instruct, an 11.2B-parameter instruction model from speakleash. GGUF is the file format that local runners such as llama.cpp and Ollama read. The copies come in five precisions, so you can trade file size against quality.

Who it suits: People who want a chat model on their own machine and work in Polish or a neighbouring language. The card lists Polish, English, Albanian, Bosnian, Bulgarian, Croatian, Czech and Danish.

Title: Who uses it, and how
Industry | Who uses it | How they use it | Example
Education | Students and teachers | Ask questions and practise in one of the listed languages | A student asks for a short explanation in Polish
Content | Writers and editors | Draft and reword text without sending it to a cloud service | An editor asks for a shorter version of a paragraph
Software | Developers | Test a local instruct model through llama.cpp or Ollama | A developer points an app at a local copy
Localization | Teams in central and southern Europe | Try one model across several of these languages | A team compares answers to one question in two languages

Skip it if you need a model with a large published feature list, because this page of the card names only the formats and runners.

Where to use it:

• llama.cpp
• Ollama
• Windows, macOS or Linux for downloading the files, which needs Python

Use case | Who | What they get back
Local chat | Individuals | Text replies from a model that stays on your computer
Drafting | Writers | A first draft or rewrite in a supported language
App testing | Developers | A local model to try an app against

How to set it up: You need Python and pip to download the files, plus llama.cpp or Ollama to run them. Hugging Face estimates about 5.6 GB at 4-bit.

• Python and pip
• llama.cpp or Ollama
• Free disk space for the quant you pick

1. Install the Hugging Face hub library.
2. Download the repository.
3. Open one of the GGUF files in llama.cpp or Ollama.

pip install -U huggingface_hub
hf download speakleash/Bielik-11B-v3.1-Instruct-GGUF

Title: The five precisions
Quant | When to pick it
bf16 | The full-precision copy, the largest of the set
Q8_0 | Near full quality, with a big file
Q6_K | High quality at a smaller size
Q5_K_M | A middle choice between size and quality
Q4_K_M | The smallest listed, near 5.6 GB by Hugging Face's estimate

How to use it: Load a GGUF file in llama.cpp or Ollama and chat with it as an instruction model. The card gives no command of its own, so follow your runner's own steps for loading a local GGUF file.

Tool | What it does
llama.cpp | Runs GGUF files directly on your computer
Ollama | Runs local models and keeps them in a library on your machine

Example: You open the model and ask a question in Polish, such as a short explanation of a school topic. The reply should come back as text in the same language. Check anything important, since the card shares no test results.

For newcomers and experts: If you are new, pick the Q4_K_M file, because it is the smallest and the easiest to try. Load it in your runner and ask one short question in a language you know.

For people in the field, this is a quantized release of speakleash/Bielik-11B-v3.1-Instruct, with five levels from bf16 down to Q4_K_M. Lower levels shrink the file and use less memory, at some cost in accuracy. The card gives no benchmark figures for the quants.

Download the GGUF repository
Pick a quant that fits your memory
Load it in llama.cpp or Ollama
Chat with the instruct model

Good to know:

• License: Apache 2.0
• Cost to run: none named, you use your own computer
• Data it reaches: it downloads files from Hugging Face, and the card names nothing else
• Not for: none named, though the card gives no benchmark scores to rely on

On Hugging Face, 13 November 2026

• Parameters: 11.2B
• Weights: about 5.6 GB at 4-bit (estimate)
• Formats: GGUF
• Downloads in the last 30 days: 1,872
• Likes: 8
• License: apache-2.0
• Languages: pl, en, sq, bs, bg, hr, cs, da
• Based on: speakleash/Bielik-11B-v3.1-Instruct
• Tags: gguf, bielik, quantized, multilingual, conversational

Links

• Model: https://huggingface.co/speakleash/Bielik-11B-v3.1-Instruct-GGUF
• Files: https://huggingface.co/speakleash/Bielik-11B-v3.1-Instruct-GGUF/tree/main

Get the model

Download its files with Hugging Face's command line:

pip install -U huggingface_hub
hf download speakleash/Bielik-11B-v3.1-Instruct-GGUF

Then follow the usage in its model card. Nothing in this kit runs by itself.