# Can the Arc A310 run LFM3-Omni-2B?

> Yes. The Arc A310 runs LFM3-Omni-2B fully in VRAM at Q8_0 (about 2.7 GB of 4.0 GB), at roughly 30 tokens/s.

Arc A310: 4 GB (4 GB usable), 124 GB/s. Estimates at an 8K context.

## Every quantisation

| Quant | Quality | ≈ GB | Fit | Est. tok/s |
|---|---|---|---|---|
| Q2_K | Heavy loss | 1.57 | fits | 96 |
| Q3_K_M | Noticeable loss | 1.91 | fits | 64 |
| Q4_K_M | Small loss | 1.95 | fits | 52 |
| Q5_K_M | Near-lossless | 2.38 | fits | 44 |
| Q6_K | Near-lossless | 2.62 | fits | 38 |
| Q8_0 (pick) | Lossless | 2.69 | fits | 30 |
| F16 | Lossless | 4.27 | offload | 15 |

## Run it

From the Ollama library (ollama):

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

Q8_0 straight from Hugging Face (ollama):

```bash
ollama run hf.co/LiquidAI/LFM3-Omni-2B-GGUF:Q8_0
```

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

```bash
llama-server -hf LiquidAI/LFM3-Omni-2B-GGUF:Q8_0
```

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-Omni-2B-GGUF:Q8_0 --host 0.0.0.0 --port 8080 -ngl 99
```

Download with the lms CLI (lmstudio):

```bash
lms get LiquidAI/LFM3-Omni-2B-GGUF@Q8_0
```


## Omni models that fit comfortably

- [LiquidAI/d1-Omni-2B](https://selfhost.directory/models/liquidai/d1-omni-2b/on/arc-a310): Q8_0 ≈2.69 GB, est. 30 tok/s
- [LiquidAI/LFM3-Audio-2B](https://selfhost.directory/models/liquidai/lfm3-audio-2b/on/arc-a310): Q8_0 ≈2.82 GB, est. 32 tok/s
- [deepseek-ai/Janus-1.3B](https://selfhost.directory/models/deepseek-ai/janus-1.3b/on/arc-a310): Q6_K ≈3.68 GB, est. 40 tok/s
- [deepseek-ai/JanusFlow-1.3B](https://selfhost.directory/models/deepseek-ai/janusflow-1.3b/on/arc-a310): Q6_K ≈3.65 GB, est. 40 tok/s
- [lmms-lab/Aero-1-Audio](https://selfhost.directory/models/lmms-lab/aero-1-audio/on/arc-a310): Q8_0 ≈3.23 GB, est. 27 tok/s
- [Menlo/mini-Ichigo-llama3.2-3B-s-instruct](https://selfhost.directory/models/menlo/mini-ichigo-llama3.2-3b-s-instruct/on/arc-a310): Q5_K_M ≈3.65 GB, est. 30 tok/s

## Other GPUs that run it

- [GeForce RTX 4060](https://selfhost.directory/models/liquidai/lfm3-omni-2b/on/rtx-4060): F16, est. 35 tok/s
- [Apple M4 (16 GB)](https://selfhost.directory/models/liquidai/lfm3-omni-2b/on/apple-m4-16gb): F16, est. 15 tok/s
- [Radeon RX 6700 XT](https://selfhost.directory/models/liquidai/lfm3-omni-2b/on/rx-6700-xt): F16, est. 49 tok/s
- [GeForce RTX 3060 12GB](https://selfhost.directory/models/liquidai/lfm3-omni-2b/on/rtx-3060): F16, est. 46 tok/s
- [Arc B580](https://selfhost.directory/models/liquidai/lfm3-omni-2b/on/arc-b580): F16, est. 58 tok/s
- [Tesla T4](https://selfhost.directory/models/liquidai/lfm3-omni-2b/on/tesla-t4): F16, est. 41 tok/s
- [Arc A770 16GB](https://selfhost.directory/models/liquidai/lfm3-omni-2b/on/arc-a770): F16, est. 71 tok/s
- [Radeon RX 7800 XT](https://selfhost.directory/models/liquidai/lfm3-omni-2b/on/rx-7800-xt): F16, est. 79 tok/s

---
Model: https://selfhost.directory/models/liquidai/lfm3-omni-2b. GPU: https://selfhost.directory/models/gpus/arc-a310. Source: https://selfhost.directory/models/liquidai/lfm3-omni-2b/on/arc-a310