Yes. The Arc B580 runs ZwZ2-4B fully in VRAM at FP16 / BF16 (about 9.7 GB of 12 GB), at roughly 28 tokens/s.
- best quant
- FP16 / BF16
- needs
- 9.7 GB
- usable
- 12 GB
- est. speed
- ~28 tok/s
Estimates for an 8K context; offload assumes 32 GB of system RAM.Try other contexts on the ZwZ2-4B page →
Every quantisation of ZwZ2-4B on the Arc B580
| Quant | Memory | Fit | Speed |
|---|---|---|---|
| Q2_K | 3.2 GB | fits | ~168 tok/s |
| Q3_K_M | 3.5 GB | fits | ~113 tok/s |
| Q4_K_M | 4.2 GB | fits | ~91 tok/s |
| Q5_K_M | 4.7 GB | fits | ~78 tok/s |
| Q6_K | 5.1 GB | fits | ~67 tok/s |
| Q8_0 | 5.9 GB | fits | ~52 tok/s |
| FP16 / BF16 pick | 9.7 GB | fits | ~28 tok/s |
Speed is a bandwidth-based estimate for generating text (456 GB/s); prompt processing and long contexts are slower.
Measured speed on the Arc B580
No community reports for this pair yet.
Report your own speed →Run ZwZ2-4B on the Arc B580
Using FP16 / BF16, the best fit for Arc B580.
ollama run hf.co/inclusionAI/ZwZ2-4B-GGUF:F16Ollama’s ROCm/CUDA builds don’t cover this GPU. Recent Ollama releases use Vulkan when its Vulkan backend and driver are installed (older ones needed OLLAMA_VULKAN=1); llama.cpp or LM Studio with Vulkan are the safer bet.
llama-server -hf inclusionAI/ZwZ2-4B-GGUF:F16Needs a Vulkan build of llama.cpp. Downloads the GGUF on first run.
docker run -d --name llama --device /dev/dri -p 8080:8080 \
-v ~/.cache/llama.cpp:/root/.cache/llama.cpp \
ghcr.io/ggml-org/llama.cpp:server-vulkan \
-hf inclusionAI/ZwZ2-4B-GGUF:F16 --host 0.0.0.0 --port 8080 -ngl 99Server and web UI on port 8080.
lms get inclusionAI/ZwZ2-4B-GGUF@F16Then load it from the LM Studio app or with `lms load`.
from transformers import pipeline
pipe = pipeline("image-text-to-text", model="inclusionAI/ZwZ2-4B", dtype="auto", device_map="auto")
messages = [
{
"role": "user",
"content": [
{"type": "image", "url": "https://huggingface.co/datasets/huggingface/documentation-images/resolve/main/bee.jpg"},
{"type": "text", "text": "Describe this image."},
],
}
]
out = pipe(text=messages, max_new_tokens=256)
print(out[0]["generated_text"][-1]["content"])Vision models that fit the Arc B580 comfortably
All that fit →| Model | Params | Best quant | Memory | Speed |
|---|---|---|---|---|
| 4.7B | Q8_0 | ≈5.9 GBFits in memory | ~51 tok/s | |
| 4.3B | Q8_0 | ≈5.2 GBFits in memory | ~55 tok/s | |
| 4.4B | Q8_0 | ≈5.8 GBFits in memory | ~53 tok/s | |
| 4.3B | Q8_0 | ≈5.6 GBFits in memory | ~55 tok/s | |
| 3.8B | Q8_0 | ≈4.0 GBFits in memory | ~63 tok/s | |
| 5B | Q8_0 | ≈5.6 GBFits in memory | ~47 tok/s |
Other GPUs that run ZwZ2-4B
Frequently asked questions
- Can the Arc B580 run ZwZ2-4B?
- Yes. The Arc B580 runs ZwZ2-4B fully in VRAM at FP16 / BF16 (about 9.7 GB of 12 GB), at roughly 28 tokens/s.
- How fast does ZwZ2-4B run on the Arc B580?
- An estimate from memory bandwidth gives about 28 tokens/s at FP16 / BF16.
- Which quantisation of ZwZ2-4B should I use on the Arc B580?
- FP16 / BF16: the largest one that fits. Expected quality: lossless compared with the full-precision weights.
Memory and speed are estimates from the model's size and the card's memory bandwidth; real numbers depend on the runtime, drivers and context length. Community figures are medians of user reports.