selfhost.directory

The index of self-hostable projects — ranked, categorized and served with live updates.
Own your data and stop giving it to big tech.

LiquidAI logo

Can the Arc A770 16GB run LFM3-Audio-2B?

LFM3-Audio-2B·2B·Arc A770 16GB·16 GB VRAM

Saved in this browser, used on every model page.
Yes

Yes. The Arc A770 16GB runs LFM3-Audio-2B fully in VRAM at FP16 / BF16 (about 4.4 GB of 16 GB), at roughly 77 tokens/s.

best quant
FP16 / BF16
needs
4.4 GB
usable
16 GB
est. speed
~77 tok/s

Estimates for an 8K context; offload assumes 32 GB of system RAM.Try other contexts on the LFM3-Audio-2B page →

Every quantisation of LFM3-Audio-2B on the Arc A770 16GB

QuantQualityMemoryFitSpeed
Q2_K Heavy loss1.7 GBfits~468 tok/s
Q3_K_M Noticeable loss1.9 GBfits~315 tok/s
Q4_K_M Small loss2.1 GBfits~254 tok/s
Q5_K_M Near-lossless2.2 GBfits~216 tok/s
Q6_K Near-lossless2.4 GBfits~187 tok/s
Q8_0 Lossless2.8 GBfits~145 tok/s
FP16 / BF16 pickLossless4.4 GBfits~77 tok/s

Speed is a bandwidth-based estimate for generating text (560 GB/s); prompt processing and long contexts are slower.

Measured speed on the Arc A770 16GB

No community reports for this pair yet.

Report your own speed →

Run LFM3-Audio-2B on the Arc A770 16GB

Using FP16 / BF16, the best fit for Arc A770 16GB.

Ollama · From the Ollama librarydocs
ollama run lfm3-audio:2b

The library build picks its own default quantisation. Ollama’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.

Ollama · F16 straight from Hugging Facedocs
ollama run hf.co/LiquidAI/LFM3-Audio-2B-GGUF:F16

Ollama’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.cpp · OpenAI-compatible server with a web UIdocs
llama-server -hf LiquidAI/LFM3-Audio-2B-GGUF:F16

Needs a Vulkan build of llama.cpp. Downloads the GGUF on first run.

llama.cpp · Docker with Vulkandocs
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

Server and web UI on port 8080.

LM Studio · Download with the lms CLIdocs
lms get LiquidAI/LFM3-Audio-2B-GGUF@F16

Then load it from the LM Studio app or with `lms load`.

Omni models that fit the Arc A770 16GB comfortably

All that fit →
ModelParamsBest quantMemorySpeedLicence
LiquidAI logoLFM3-Omni-2BLiquidAI
2.2BQ8_0≈2.7 GBFits in memory~134 tok/sConditions
LiquidAI logod1-Omni-2BLiquidAI
2.2BQ8_0≈2.7 GBFits in memory~134 tok/sConditions
deepseek-ai logoJanus-1.3Bdeepseek-ai
2.1BQ8_0≈4.2 GBFits in memory~139 tok/sCommercial OK
5.1BQ8_0≈5.7 GBFits in memory~57 tok/sCommercial OK
deepseek-ai logoJanusFlow-1.3Bdeepseek-ai
2BQ8_0≈4.1 GBFits in memory~142 tok/sCommercial OK
OmniGen2 logoOmniGen2OmniGen2
4BQ8_0≈7.7 GBFits in memory~73 tok/sCommercial OK

Other GPUs that run LFM3-Audio-2B

Frequently asked questions

Can the Arc A770 16GB run LFM3-Audio-2B?
Yes. The Arc A770 16GB runs LFM3-Audio-2B fully in VRAM at FP16 / BF16 (about 4.4 GB of 16 GB), at roughly 77 tokens/s.
How fast does LFM3-Audio-2B run on the Arc A770 16GB?
An estimate from memory bandwidth gives about 77 tokens/s at FP16 / BF16.
Which quantisation of LFM3-Audio-2B should I use on the Arc A770 16GB?
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.