DeepSeek LLM 67B Chat on NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip

Can the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip run DeepSeek LLM 67B Chat?

Yes, it runs great

Runs great. At Q8_0 it needs about 86.2 GB of 252 GB usable, with an estimated speed of 65.8 tok/s.

Best quant: Q8_0
Generation speed: 65.8 tok/s
Prompt processing: 855 tok/s
First token (1K prompt): 1.2 s
Largest safe context: 4K

Best app to run it on the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip

On Windows

Ollama

Command line + local API · runs on llama.cpp (ggml), MLX (Apple silicon, preview)

One command with Ollama’s own build of this model.
Serves an OpenAI-compatible API that other apps and coding tools can use.

ollama run deepseek-llm:67b

LM Studio Point-and-click: search, download and chat in one window.
llama.cpp Usually the first to support new model architectures.

On Linux

Ollama

Command line + local API · runs on llama.cpp (ggml), MLX (Apple silicon, preview)

One command with Ollama’s own build of this model.
Serves an OpenAI-compatible API that other apps and coding tools can use.

ollama run deepseek-llm:67b

vLLM The full model fits in GPU memory at 8-bit, so it can batch many requests: best for serving a team or an app.
LM Studio Point-and-click: search, download and chat in one window.

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 DeepSeek LLM 67B Chat

Quickest · Ollama

One command downloads a ready-to-run version and starts a chat.

Get Ollama at ollama.com/download (opens in a new tab) · model page (opens in a new tab)

ollama run deepseek-llm:67b

Quantized files · GGUF

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

huggingface.co/mradermacher/deepseek-llm-67b-chat-i1-GGUF (opens in a new tab)

ollama run hf.co/mradermacher/deepseek-llm-67b-chat-i1-GGUF

Original weights · Hugging Face

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

huggingface.co/deepseek-ai/deepseek-llm-67b-chat (opens in a new tab)

No sign-in needed.

Links checked 3 days ago against Hugging Face, Ollama and GitHub.

How it’s rated

Quality 40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking.: How capable the model itself is at Q8_0, 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 NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip: 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.	18.0 / 45
Memory fitFull marks when everything sits in GPU memory; fewer when part spills to system RAM.	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.	7.9 / 15
Match score	65.9 / 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 NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip

Quant	Memory needed	Available	Speed	Runs	Quality	Download
`Q2_K`	41.8 GB	252 GB	212 tok/s	Runs great	28Basic quality (D) · 28 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 56/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q3_K_M`	47.8 GB	252 GB	167 tok/s	Runs great	33Basic quality (D) · 33 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q4_K_M`	55.9 GB	252 GB	129 tok/s	Runs great	38Basic quality (D) · 38 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`	62.6 GB	252 GB	107 tok/s	Runs great	38Basic quality (D) · 38 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`	70.1 GB	252 GB	89.7 tok/s	Runs great	39Basic quality (D) · 39 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking.	GGUF files (opens in a new tab)
`Q8_0`	86.2 GB	252 GB	65.8 tok/s	Runs great	40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking.	GGUF files (opens in a new tab)
`F16`	150 GB	252 GB	30.3 tok/s	Runs great	40Fair quality (C) · 40 of 100: how capable the model is at F16, whatever it runs on. Match score 69/100 (quality, fit, speed and memory use) sets the ranking.	Original weights (opens in a new tab)

Memory figures assume 4K of context. Longer conversations or documents grow the KV cache.

Other hardware that runs it

38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA RTX PRO 6000 Blackwell Q4_K_M · 32.6 tok/s · Runs great
28Basic quality (D) · 28 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 66/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GeForce RTX 5090 Q2_K · 53.6 tok/s · Runs well (tight fit)
33Basic quality (D) · 33 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 63/100 (quality, fit, speed and memory use) sets the ranking. Apple M4 Max (64 GB) Q3_K_M · 11.1 tok/s · Runs well
28Basic quality (D) · 28 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking. Apple M1 Max (64 GB) Q2_K · 10.3 tok/s · Runs well
38Basic quality (D) · 38 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking. Apple M2 Ultra (192 GB) Q4_K_M · 12.5 tok/s · Runs well
33Basic quality (D) · 33 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 58/100 (quality, fit, speed and memory use) sets the ranking. Apple M4 Max (128 GB) Q3_K_M · 11.1 tok/s · Runs well

Quick answers

How much memory does DeepSeek LLM 67B Chat need on the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip 252GB?

At Q8_0 it needs about 86.2 GB. The NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip 252GB has 252 GB usable.

How fast is DeepSeek LLM 67B Chat on the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip 252GB?

About 65.8 tok/s at Q8_0, an estimate from the hardware’s memory bandwidth and the size of the model.

Which quantization of DeepSeek LLM 67B Chat should I use on the NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip 252GB?

Q8_0. Of the versions that fit, it is the one that runs best; when two run equally well, the check prefers the higher-rated one.

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.