Chirp Chirp! 🐦 We are introducing Ornith-1.5, a major step toward building foundation models through end-to-end self-improvement.
Bielik-11B-v3.1-Instruct
speakleash·Instruct·released 26 Oct 2026
ollama run hf.co/speakleash/Bielik-11B-v3.1-Instruct-GGUF:Q4_K_MBielik-11B-v3.1-Instruct is a generative text model featuring 11 billion parameters. It is an update to the Bielik 11B v3 line that expands multilingual continual pre-training and strengthens post-training for instruction following, tool use, and reasoning.
How much VRAM does Bielik-11B-v3.1-Instruct need?
At Q4_K_M Bielik-11B-v3.1-Instruct needs about 8.3 GB with an 8K context (8.3 GB at Q4, 24 GB at full precision). Pick your GPU to see what fits and how fast it runs.
Needs: ≈8.3 GB · Q4_K_M · 8K context — Pick a GPU to check the fit.
Advanced: context 8K · KV cache 16-bit
Smallest GPU for each quant and context (KV cache 16-bit)
| Quantisation | 4K | 8K | 16K | 32K |
|---|---|---|---|---|
| Q3_K_M | ||||
| Q4_K_M | ||||
| Q5_K_M | ||||
| Q6_K | ||||
| Q8_0 | ||||
| FP16 |
≤16 GB card24–32 GB48–96 GBmore
| Quantisation | Total |
|---|---|
| Q2_KHeavy lossSmallest, noticeably worse quality | ≈5.3 GB |
| Q3_K_MNoticeable lossSmall, some quality loss | ≈7.0 GB |
| Q4_K_MrecommendedSmall lossRecommended default for local use | ≈8.3 GB |
| Q5_K_MNear-losslessNear-lossless for most tasks | ≈9.4 GB |
| Q6_KNear-losslessPractically lossless | ≈11 GB |
| Q8_0LosslessLossless in practice, twice the size of Q4 | ≈13 GB |
| FP16 / BF16LosslessFull precision, as released | ≈24 GB |
Estimates for 8K tokens of context: weights + KV cache + runtime overhead. Speed assumes memory-bound decoding at your GPU’s bandwidth. Pick a GPU above and the fit column checks every row against it. Weights marked “file” use the real GGUF download size; “est.” rows are computed from the parameter count.
How to pick a quant
Take the biggest quant that fits with room left for context. Q4_K_M is the usual sweet spot. Below Q4 quality drops fast, above Q6 you mostly pay memory for nothing. Big models (70B+, large MoE) tolerate low quants better than small ones.
- LosslessF16, Q8_0
- Near-losslessQ6_K, Q5_K_M
- Small lossQ5_K_S, Q4_K_M, IQ4_NL, IQ4_XS, Q4_K_S
- Noticeable lossQ4_0, Q3_K_L, Q3_K_M, IQ3_M
- Heavy lossIQ3_XXS, Q3_K_S, IQ2_M, Q2_K
Typical quality loss, not measured per model. Tiers come from KL divergence against FP16 on Llama 3 8B: llama.cpp perplexity README: LLaMA 3 8B scoreboard (KLD and ΔPPL vs FP16, Wikitext-2); llama.cpp quantize tool: ΔPPL per quant type @ Llama-3-8B
How much context fits?
At 8K context, Q4_K_M needs ≈8.3 GB(6.3 GB weights + 1.2 GB KV cache + 0.8 GB overhead). Pick a GPU above to see how much context fits.
Weights + overheadKV cache
Can the Apple M4 (16 GB) run Bielik-11B-v3.1-Instruct? →
Speed vs quality per quant
Smaller quants run faster and fit in less memory, but lose quality.
Fits in VRAMTight fitCPU offload✕ won't runQuality is typical KL divergence vs FP16, not measured for this model.
Real-world speed of Bielik-11B-v3.1-Instruct
Tokens per second measured by people running it at home. A GPU shows a median once 3 reports agree on the same setup. Reports far from our estimate wait for a review. Measure yours with llama-bench →
No medians yet. Be one of the first 3 to report a setup.
All reports
Run Bielik-11B-v3.1-Instruct locally
--quantQ4_K_Mdefault; pick your GPU to tailor itspeakleash/Bielik-11B-v3.1-Instruct-GGUF
ollama run hf.co/speakleash/Bielik-11B-v3.1-Instruct-GGUF:Q4_K_M llama-server -hf speakleash/Bielik-11B-v3.1-Instruct-GGUF:Q4_K_MDownloads the GGUF on first run. Installed through llama.app? The same flags work with `llama serve`.
lms get speakleash/Bielik-11B-v3.1-Instruct-GGUF@Q4_K_MThen load it from the LM Studio app or with `lms load`.
pip install mlx-lm
mlx_lm.generate --model speakleash/Bielik-11B-v3.1-Instruct-MLX-4bit --prompt "Hello"Use `mlx_lm.chat` with the same --model for an interactive chat.
mlx_lm.server --model speakleash/Bielik-11B-v3.1-Instruct-MLX-4bit --port 8080API on http://localhost:8080/v1.
vllm serve speakleash/Bielik-11B-v3.1-InstructGated on Hugging Face: accept the licence on the model page and export HF_TOKEN first.
vllm serve speakleash/Bielik-11B-v3.1-Instruct-AWQservices:
vllm:
image: vllm/vllm-openai:latest
command: speakleash/Bielik-11B-v3.1-Instruct
ports:
- "8000:8000"
ipc: host
volumes:
- ~/.cache/huggingface:/root/.cache/huggingface
environment:
- HF_TOKEN=${HF_TOKEN}
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: all
capabilities: [gpu]API on port 8000.
sglang serve speakleash/Bielik-11B-v3.1-Instruct --host 0.0.0.0 --port 30000Gated on Hugging Face: accept the licence on the model page and export HF_TOKEN first.
services:
sglang:
image: lmsysorg/sglang:latest
command: sglang serve speakleash/Bielik-11B-v3.1-Instruct --host 0.0.0.0 --port 30000
ports:
- "30000:30000"
ipc: host
volumes:
- ~/.cache/huggingface:/root/.cache/huggingface
environment:
- HF_TOKEN=${HF_TOKEN}
deploy:
resources:
reservations:
devices:
- driver: nvidia
count: all
capabilities: [gpu]API on port 30000.
from transformers import pipeline
pipe = pipeline("text-generation", model="speakleash/Bielik-11B-v3.1-Instruct", dtype="auto", device_map="auto")
messages = [{"role": "user", "content": "Explain self-hosting in one paragraph."}]
out = pipe(messages, max_new_tokens=512)
print(out[0]["generated_text"][-1]["content"])Gated on Hugging Face: accept the licence on the model page and export HF_TOKEN first.
Quantised versions and downloads
| Repository | Format | Quant | Size |
|---|---|---|---|
| speakleash/Bielik-11B-v3.1-Instruct-NVFP4A16 | Safetensors | NVFP4 | — |
| speakleash/Bielik-11B-v3.1-Instruct-NVFP4 | Safetensors | NVFP4 | — |
| speakleash/Bielik-11B-v3.1-Instruct-FP8-Dynamic | Safetensors | FP8 | — |
| speakleash/Bielik-11B-v3.1-Instruct-GGUF | GGUF | Q4_K_M | 6.7 GB |
| speakleash/Bielik-11B-v3.1-Instruct-GGUF | GGUF | Q5_K_M | 7.9 GB |
| speakleash/Bielik-11B-v3.1-Instruct-GGUF | GGUF | Q6_K | 9.2 GB |
| speakleash/Bielik-11B-v3.1-Instruct-GGUF | GGUF | Q8_0 | 12 GB |
| speakleash/Bielik-11B-v3.1-Instruct-GGUF | GGUF | BF16 | 22 GB |
| speakleash/Bielik-11B-v3.1-Instruct-MLX-4bit | MLX | 4BIT | — |
| speakleash/Bielik-11B-v3.1-Instruct-MLX-8bit | MLX | 8BIT | — |
About Bielik-11B-v3.1-Instruct
Bielik-11B-v3.1-Instruct is a generative text model featuring 11 billion parameters. It is an update to the Bielik 11B v3 line that expands multilingual continual pre-training and strengthens post-training for instruction following, tool use, and reasoning. Forementioned model stands as a testament to the unique collaboration between the open-science/open-source project SpeakLeash and the High Performance Computing (HPC) center: ACC Cyfronet AGH. Developed and trained on multilingual text corpora covering 56 languages, with an emphasis on European languages and Polish, which have been cherry-picked and processed by the SpeakLeash team, this endeavor leverages Polish large-scale computing infrastructure of the ACC Cyfronet AGH HPC center. The creation and training of Bielik-11B-v3.1-Instruct was propelled by resources provided within the PLGrid environment under computational grant number [PLG/2026/019559], conducted on the Athena and Helios supercomputers, and within the Gaia AI Factory, funded by the European Union through the EuroHPC Joint Undertaking. As a result, the model exhibits an exceptional ability to understand and process the Polish language and other European languages, providing accurate responses and performing a variety of linguistic tasks with high precision. 🗣️ Chat: https://chat.bielik.ai/
Full model card on Hugging Face →Alternatives to Bielik-11B-v3.1-Instruct
All alternatives →Aloha! 🌺 Today, we are releasing Ornith-1.0, a self-improving family of open-source models for agentic coding.
Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models.
Bielik-11B-v3.0-Instruct is a generative text model featuring 11 billion parameters. It is an instruct fine-tuned version of the Bielik-11B-v3-Base-20250730.
A compact 9B reasoning model distilled from DeepSeek-V4-Pro in Max Effect mode
Qwen3 is the latest generation of large language models in Qwen series, offering a comprehensive suite of dense and mixture-of-experts (MoE) models.
Bielik-11B-v3.1-Instruct FAQ
- How much VRAM does Bielik-11B-v3.1-Instruct need?
- About 8.3 GB at Q4_K_M with an 8K context. Q8_0 needs about 13 GB and full precision about 24 GB. Longer contexts need more memory for the KV cache.
- Can Bielik-11B-v3.1-Instruct run on the RTX 3060 12GB?
- Yes. It fits at Q6_K (about 11 GB), at an estimated 22 tokens/s. It fits comfortably on the Apple M4 (16 GB), Radeon RX 6700 XT, Arc B580.
- Can I use Bielik-11B-v3.1-Instruct 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 Bielik-11B-v3.1-Instruct?
- 32,768 tokens (32K). The full window needs far more memory than an 8K context.
- How do I run Bielik-11B-v3.1-Instruct with Ollama or llama.cpp?
- With Ollama: ollama run hf.co/speakleash/Bielik-11B-v3.1-Instruct-GGUF:Q4_K_M. With llama.cpp: llama-server -hf speakleash/Bielik-11B-v3.1-Instruct-GGUF:Q4_K_M, which starts an OpenAI-compatible server.
Rate this model
Data from Hugging Face, the model card and public leaderboards. Memory and speed figures are estimates; real numbers depend on the runtime, context length and drivers. Always check the licence yourself before commercial use.
0 comments