Llama 3.2 90B Vision Instruct on NVIDIA B200

Can the NVIDIA B200 run Llama 3.2 90B Vision Instruct?

Yes, it runs great

Runs great. At Q8_0 it needs about 107 GB of 180 GB usable, with an estimated speed of 56.4 tok/s.

Best quant: Q8_0
Generation speed: 56.4 tok/s
Prompt processing: 1,300 tok/s
First token (1K prompt): 0.8 s
Largest safe context: 128K

Best app to run it on the NVIDIA B200

On Linux

vLLM

Inference server · runs on 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.

vllm serve meta-llama/Llama-3.2-90B-Vision-Instruct

Ollama One command with Ollama’s own build of this model.
SGLang The full model fits in GPU memory at 8-bit, so it can batch many requests: best for serving a team or an app.

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 llama3.2-vision:90b-instruct-q8_0

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.
SGLang The full model fits in GPU memory at 8-bit, so it can batch many requests: best for serving a team or an app.

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 Llama 3.2 90B Vision Instruct

Recommended for the NVIDIA B200: Q8_0. Every other quantization has its own download link in the table below.

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 llama3.2-vision:90b-instruct-q8_0

Original weights · Hugging Face

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

huggingface.co/meta-llama/Llama-3.2-90B-Vision-Instruct (opens in a new tab)

Free Hugging Face account needed: request access on the model page, and Meta approves it (usually within minutes to a few hours).

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

How it’s rated

Quality 66Good quality (B) · 66 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 83/100 (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 B200: 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.	29.7 / 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.	13.7 / 15
Match score	83.4 / 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 B200

Quant	Memory needed	Available	Speed	Runs	Quality	Download
`Q2_K`	48.4 GB	180 GB	182 tok/s	Runs great	46Fair quality (C) · 46 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.	All files (opens in a new tab)
`Q3_K_M`	56.4 GB	180 GB	143 tok/s	Runs great	55Good quality (B) · 55 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 72/100 (quality, fit, speed and memory use) sets the ranking.	All files (opens in a new tab)
`Q4_K_M`	67.0 GB	180 GB	111 tok/s	Runs great	62Good quality (B) · 62 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 76/100 (quality, fit, speed and memory use) sets the ranking.
`Q5_K_M`	75.9 GB	180 GB	91.6 tok/s	Runs great	64Good quality (B) · 64 of 100: how capable the model is at Q5_K_M, whatever it runs on. Match score 78/100 (quality, fit, speed and memory use) sets the ranking.	All files (opens in a new tab)
`Q6_K`	85.6 GB	180 GB	76.9 tok/s	Runs great	65Good quality (B) · 65 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking.	All files (opens in a new tab)
`Q8_0`	107 GB	180 GB	56.4 tok/s	Runs great	66Good quality (B) · 66 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 83/100 (quality, fit, speed and memory use) sets the ranking.
`F16`	190 GB	180 GB	3.3 tok/s	Slowoffloads to RAM	66Good quality (B) · 66 of 100: how capable the model is at F16, whatever it runs on. Match score 59/100 (quality, fit, speed and memory use) sets the ranking.	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

62Good quality (B) · 62 of 100: how capable the model is at Q4_K_M, whatever it runs on. Match score 80/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA RTX PRO 6000 Blackwell Q4_K_M · 24.8 tok/s · Runs well
55Good quality (B) · 55 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. Apple M2 Ultra (192 GB) Q3_K_M · 12.3 tok/s · Runs well
46Fair quality (C) · 46 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 67/100 (quality, fit, speed and memory use) sets the ranking. Apple M4 Max (64 GB) Q2_K · 10.7 tok/s · Runs well (tight fit)
55Good quality (B) · 55 of 100: how capable the model is at Q3_K_M, whatever it runs on. Match score 65/100 (quality, fit, speed and memory use) sets the ranking. Apple M3 Ultra (256 GB) Q3_K_M · 12.6 tok/s · Runs well
46Fair quality (C) · 46 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 64/100 (quality, fit, speed and memory use) sets the ranking. Apple M4 Max (128 GB) Q2_K · 10.7 tok/s · Runs well
55Good quality (B) · 55 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 M3 Ultra (512 GB) Q3_K_M · 12.6 tok/s · Runs well

Quick answers

How much memory does Llama 3.2 90B Vision Instruct need on the NVIDIA B200 180GB?

At Q8_0 it needs about 107 GB. The NVIDIA B200 180GB has 180 GB usable.

How fast is Llama 3.2 90B Vision Instruct on the NVIDIA B200 180GB?

About 56.4 tok/s at Q8_0, an estimate from the hardware’s memory bandwidth and the size of the model.

Which quantization of Llama 3.2 90B Vision Instruct should I use on the NVIDIA B200 180GB?

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.