on

Can the Apple M3 Ultra (512 GB) run LightOnOCR-3 4B?

48Fair quality (C) · 48 of 100: how capable the model is at F16, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking.

Yes, it runs great

Runs great. At F16 it needs about 40.7 GB of 384 GB usable, with an estimated speed of 44.7 tok/s.

Best quant
F16
Generation speed
44.7 tok/s
Prompt processing
3,878 tok/s
First token (1K prompt)
0.3 s
Largest safe context
256K

Best app to run it on the Apple M3 Ultra (512 GB)

On macOS

LM Studio

Desktop app · runs on llama.cpp, MLX

  • Point-and-click: search, download and chat in one window.

Open LightOnOCR-3 4B in LM Studio

  • llama.cpp Usually the first to support new model architectures.
  • Ollama Runs the Hugging Face GGUF with ollama run hf.co/…

How we choose: the files published for this model, your operating system and GPU, whether it fits fully in memory, and published speed tests (MLX vs llama.cpp on Apple silicon, vLLM vs llama.cpp vs Ollama on NVIDIA). Details.

Download LightOnOCR-3 4B

Recommended for the Apple M3 Ultra (512 GB): F16. 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 F16 file (opens in a new tab)

huggingface.co/mradermacher/LightOnOCR-3-4B-GGUF (opens in a new tab)

ollama run hf.co/mradermacher/LightOnOCR-3-4B-GGUF:F16

Original weights · Hugging Face

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

huggingface.co/lightonai/LightOnOCR-3-4B (opens in a new tab)

No sign-in needed.

Official repo · GitHub

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

github.com/lightonai/LightOnOCR (opens in a new tab)

LightOn's official LightOnOCR repo: a client, CLI and viewer for vLLM-served models, plus scripts that reproduce the benchmark scores.

git clone https://github.com/lightonai/LightOnOCR

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

How it’s rated

Quality 48Fair quality (C) · 48 of 100: how capable the model is at F16, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking.
How capable the model itself is at F16, 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 great
How well it runs on the Apple M3 Ultra (512 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.21.6 / 45
Memory fitFull marks when the model fits in the share of unified memory the GPU can use; fewer when it goes past that limit.25.0 / 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.2.4 / 15
Match score64.0 / 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 Apple M3 Ultra (512 GB)

Quant Memory needed Available Speed Runs Quality Download
Q2_K 33.4 GB 384 GB 313 tok/s Runs great 33Basic quality (D) · 33 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 57/100 (quality, fit, speed and memory use) sets the ranking. GGUF (opens in a new tab)
Q3_K_M 33.8 GB 384 GB 247 tok/s Runs great 40Fair quality (C) · 40 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. GGUF (opens in a new tab)
Q4_K_M 34.4 GB 384 GB 191 tok/s Runs great 45Fair quality (C) · 45 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 62/100 (quality, fit, speed and memory use) sets the ranking. GGUF (opens in a new tab)
Q5_K_M 34.8 GB 384 GB 158 tok/s Runs great 46Fair quality (C) · 46 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. GGUF (opens in a new tab)
Q6_K 35.3 GB 384 GB 132 tok/s Runs great 47Fair quality (C) · 47 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. GGUF (opens in a new tab)
Q8_0 36.4 GB 384 GB 97.1 tok/s Runs great 48Fair quality (C) · 48 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. GGUF (opens in a new tab)
F16 40.7 GB 384 GB 44.7 tok/s Runs great 48Fair quality (C) · 48 of 100: how capable the model is at F16, whatever it runs on. Match score 64/100 (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?