Run speakleash/Bielik-11B-v3.0-Instruct-awq locally
speakleash/Bielik-11B-v3.0-Instruct-awq is a large instruction-tuned chat model with 11.34 billion parameters, built on the llama architecture. It is released under the apache-2.0 license and has been downloaded 535,890 times.
To run speakleash/Bielik-11B-v3.0-Instruct-awq locally at a 4,096-token context, its quantized versions need between 8.1 GB (4-bit, lowest quality) and 14.62 GB (8-bit, highest quality) of memory, weights plus KV cache and a system margin included.
For most users the best balance is 8-bit, needing about 14.62 GB. That means speakleash/Bielik-11B-v3.0-Instruct-awq fits entirely in the VRAM of a 10 GB GPU or larger, running fully on the GPU.
All quantizations
| Quant. | Bits | Quality | Weights | KV | Total | Speed~ | Verdict |
|---|---|---|---|---|---|---|---|
| 4-bit | 4.0 | Fair | 6.52 GB | 0.78 GB | 8.1 GB | 7.7 t/s | Offload |
| 8-bit | 8.0 | Excellent | 13.04 GB | 0.78 GB | 14.62 GB | 3.8 t/s | Offload |
KV cache computed from the model's exact architecture. Speed is a rough estimate bounded by memory bandwidth.
Frequently asked questions
How much VRAM do you need to run speakleash/Bielik-11B-v3.0-Instruct-awq?
You need about 8.1 GB of VRAM to run speakleash/Bielik-11B-v3.0-Instruct-awq entirely on the GPU using the 4-bit quantization (at a 4,096-token context). Smaller quantizations lower the requirement at the cost of quality.
Can I run speakleash/Bielik-11B-v3.0-Instruct-awq on an 8 GB GPU?
Partially. speakleash/Bielik-11B-v3.0-Instruct-awq only fits on an 8 GB GPU by offloading part of it to system RAM (with 8-bit), which runs but is slower.
Can I run speakleash/Bielik-11B-v3.0-Instruct-awq on a 16 GB GPU?
Yes. With 16 GB of VRAM you can run speakleash/Bielik-11B-v3.0-Instruct-awq fully on the GPU using 8-bit (about 14.62 GB).
Can I run speakleash/Bielik-11B-v3.0-Instruct-awq on a 24 GB GPU?
Yes. With 24 GB of VRAM you can run speakleash/Bielik-11B-v3.0-Instruct-awq fully on the GPU using 8-bit (about 14.62 GB).
What is the best quantization for speakleash/Bielik-11B-v3.0-Instruct-awq?
If memory allows, higher bits-per-weight means better quality. A common sweet spot is a Q4_K_M or Q5_K_M quantization, which keeps most of the quality while roughly halving the memory versus 8-bit. Pick the highest quantization that still fits in your VRAM.