# Can the Arc B580 run LFM3-Audio-2B?

> Yes. The Arc B580 runs LFM3-Audio-2B fully in VRAM at FP16 / BF16 (about 4.4 GB of 12 GB), at roughly 63 tokens/s.

Arc B580: 12 GB (12 GB usable), 456 GB/s. Estimates at an 8K context.

## Every quantisation

| Quant | Quality | ≈ GB | Fit | Est. tok/s |
|---|---|---|---|---|
| Q2_K | Heavy loss | 1.66 | fits | 381 |
| Q3_K_M | Noticeable loss | 1.9 | fits | 256 |
| Q4_K_M | Small loss | 2.08 | fits | 207 |
| Q5_K_M | Near-lossless | 2.24 | fits | 176 |
| Q6_K | Near-lossless | 2.41 | fits | 152 |
| Q8_0 | Lossless | 2.82 | fits | 118 |
| F16 (pick) | Lossless | 4.4 | fits | 63 |

## Run it

From the Ollama library (ollama):

```bash
ollama run lfm3-audio:2b
```

F16 straight from Hugging Face (ollama):

```bash
ollama run hf.co/LiquidAI/LFM3-Audio-2B-GGUF:F16
```

OpenAI-compatible server with a web UI (llama.cpp):

```bash
llama-server -hf LiquidAI/LFM3-Audio-2B-GGUF:F16
```

Docker with Vulkan (llama.cpp):

```bash
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 LiquidAI/LFM3-Audio-2B-GGUF:F16 --host 0.0.0.0 --port 8080 -ngl 99
```

Download with the lms CLI (lmstudio):

```bash
lms get LiquidAI/LFM3-Audio-2B-GGUF@F16
```


## Omni models that fit comfortably

- [LiquidAI/LFM3-Omni-2B](https://selfhost.directory/models/liquidai/lfm3-omni-2b/on/arc-b580): Q8_0 ≈2.69 GB, est. 109 tok/s
- [LiquidAI/d1-Omni-2B](https://selfhost.directory/models/liquidai/d1-omni-2b/on/arc-b580): Q8_0 ≈2.69 GB, est. 109 tok/s
- [deepseek-ai/Janus-1.3B](https://selfhost.directory/models/deepseek-ai/janus-1.3b/on/arc-b580): Q8_0 ≈4.17 GB, est. 113 tok/s
- [deepseek-ai/JanusFlow-1.3B](https://selfhost.directory/models/deepseek-ai/janusflow-1.3b/on/arc-b580): Q8_0 ≈4.13 GB, est. 115 tok/s
- [google/gemma-4-E2B-it](https://selfhost.directory/models/google/gemma-4-e2b-it/on/arc-b580): Q8_0 ≈5.71 GB, est. 46 tok/s
- [OmniGen2/OmniGen2](https://selfhost.directory/models/omnigen2/omnigen2/on/arc-b580): Q8_0 ≈7.73 GB, est. 60 tok/s

## Other GPUs that run it

- [GeForce RTX 4060](https://selfhost.directory/models/liquidai/lfm3-audio-2b/on/rtx-4060): F16, est. 37 tok/s
- [Apple M4 (16 GB)](https://selfhost.directory/models/liquidai/lfm3-audio-2b/on/apple-m4-16gb): F16, est. 16 tok/s
- [Radeon RX 6700 XT](https://selfhost.directory/models/liquidai/lfm3-audio-2b/on/rx-6700-xt): F16, est. 53 tok/s
- [GeForce RTX 3060 12GB](https://selfhost.directory/models/liquidai/lfm3-audio-2b/on/rtx-3060): F16, est. 49 tok/s
- [Tesla T4](https://selfhost.directory/models/liquidai/lfm3-audio-2b/on/tesla-t4): F16, est. 44 tok/s
- [Arc A770 16GB](https://selfhost.directory/models/liquidai/lfm3-audio-2b/on/arc-a770): F16, est. 77 tok/s
- [Radeon RX 7800 XT](https://selfhost.directory/models/liquidai/lfm3-audio-2b/on/rx-7800-xt): F16, est. 86 tok/s
- [Tesla P40](https://selfhost.directory/models/liquidai/lfm3-audio-2b/on/tesla-p40): F16, est. 48 tok/s

---
Model: https://selfhost.directory/models/liquidai/lfm3-audio-2b. GPU: https://selfhost.directory/models/gpus/arc-b580. Source: https://selfhost.directory/models/liquidai/lfm3-audio-2b/on/arc-b580