on

Can the AMD EPYC 7002 / 7003 series NAS (16 GB) run ERNIE 4.5 21B-A3B Thinking?

49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking.

Yes, it runs well (tight fit)

Runs well (tight fit). At Q2_K it needs about 10.8 GB of 12.0 GB usable, with an estimated speed of 61.7 tok/s.

Best quant
Q2_K
Generation speed
61.7 tok/s
Prompt processing
133 tok/s
First token (1K prompt)
7.7 s
Largest safe context
29K

Best app to run it on the AMD EPYC 7002 / 7003 series NAS (16 GB)

On Linux

llama.cpp

Command line + local API · runs on llama.cpp

  • Usually the first to support new model architectures.
llama-server -hf unsloth/ERNIE-4.5-21B-A3B-Thinking-GGUF:Q2_K

How we choose: the files published for this model, which apps run in a container on a NAS without a graphics card (models run on its processor), and whether the model fits fully in the memory models can use. Details.

Download ERNIE 4.5 21B-A3B Thinking

Recommended for the AMD EPYC 7002 / 7003 series NAS (16 GB): Q2_K. Every other quantization has its own download link in the table below.

Quantized files · GGUF

Works in LM Studio, llama.cpp, Jan and Ollama. No sign-in needed.

Download the Q2_K file (opens in a new tab)

huggingface.co/unsloth/ERNIE-4.5-21B-A3B-Thinking-GGUF (opens in a new tab)

ollama run hf.co/unsloth/ERNIE-4.5-21B-A3B-Thinking-GGUF:Q2_K

Original weights · Hugging Face

Full-precision files from Baidu, for fine-tuning, vLLM, Transformers.

huggingface.co/baidu/ERNIE-4.5-21B-A3B-Thinking (opens in a new tab)

No sign-in needed.

Official repo · GitHub

Reference code, inference scripts and docs from Baidu. The weights themselves download from Hugging Face.

github.com/PaddlePaddle/ERNIE (opens in a new tab)

git clone https://github.com/PaddlePaddle/ERNIE

Links checked 1 day ago against Hugging Face, Ollama and GitHub.

How it’s rated

Quality 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking.
How capable the model itself is at Q2_K, from its editorial quality tier scaled by the quantization or precision. It’s the same on any hardware that runs this variant. Scale: A+ 85+ · A 70+ · B 55+ · C 40+ · D under 40.
Runs Runs welltight fit
How well it runs on the AMD EPYC 7002 / 7003 series NAS (16 GB): speed, capped by how well it fits in memory. Runs great: 30+ tokens/s (or near the image, video or decision speed target) · Runs well: 10+ tokens/s · Slow: 3+ tokens/s · Very slow: under 3 · Won’t run: doesn’t fit even with offload. A tight fit or RAM offload caps it at Runs well; heavy offload caps it at Slow.

Match score

To order lists, each result also gets a match score out of 100 from four parts. It favors the most capable model that still runs well.

Model qualityEditorial quality tier of the model, scaled by the quantization or precision used.22.1 / 45
Memory fitFull marks when the model fits in the share of the NAS’s memory that models can use (the rest stays with the NAS system and its storage services); fewer when it goes past that limit.24.0 / 25
SpeedGeneration speed against a 40 tokens/s target for LLMs, or render time against a target for image and video.15.0 / 15
Memory useRewards using the memory you have, so a bigger, better variant beats a tiny one that leaves most memory idle.15.0 / 15
Match score76.1 / 100

The match score orders results; it is not shown as a grade. Quality and Runs are the two ratings in the tables.

Every quantization on the AMD EPYC 7002 / 7003 series NAS (16 GB)

Quant Memory needed Available Speed Runs Quality Download
Q2_K 10.8 GB 12.0 GB 61.7 tok/s Runs welltight fit 49Fair quality (C) · 49 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking. GGUF (opens in a new tab)
Q3_K_M 12.7 GB 12.0 GB 42.8 tok/s Runs wellover memory limit 58Good quality (B) · 58 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. GGUF (opens in a new tab)
Q4_K_M 15.3 GB 12.0 GB — Won't run 66Good quality (B) · 66 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 8/100 (quality, fit, speed and memory use) sets the ranking. GGUF (opens in a new tab)
Q5_K_M 17.5 GB 12.0 GB — Won't run 67Good quality (B) · 67 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. GGUF (opens in a new tab)
Q6_K 19.9 GB 12.0 GB — Won't run 69Good quality (B) · 69 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. GGUF (opens in a new tab)
Q8_0 25.2 GB 12.0 GB — Won't run 70Great quality (A) · 70 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. GGUF (opens in a new tab)
F16 45.7 GB 12.0 GB — Won't run 70Great quality (A) · 70 of 100: how capable the model is at F16, whatever it runs on. Match score 9/100 (quality, fit, speed and memory use) sets the ranking. GGUF (opens in a new tab) Original weights (opens in a new tab)

Memory figures assume 8K of context. Longer conversations or documents grow the KV cache.

Other hardware that runs it

Estimates come from published hardware specs and our open formulas, not from running the model on your machine. Real speed varies with runtime, drivers and settings. Numbers marked “Measured” come from a linked benchmark.

Was this page helpful?