Phi-3.5 Mini Instruct on NVIDIA H100 NVL

Can the NVIDIA H100 NVL run Phi-3.5 Mini Instruct?

Yes, it runs great

Runs great. At F16 it needs about 16.2 GB of 94.0 GB usable, with an estimated speed of 294 tok/s.

Best quant: F16
Generation speed: 294 tok/s
Prompt processing: 22,618 tok/s
First token (1K prompt): 0.3 s
Largest safe context: 128K

Best app to run it on the NVIDIA H100 NVL

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 microsoft/Phi-3.5-mini-instruct

Ollama One command with Ollama’s own build of this model.
LM Studio Point-and-click: search, download and chat in one window.

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 phi3.5:3.8b-mini-instruct-fp16

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

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 Phi-3.5 Mini Instruct

Recommended for the NVIDIA H100 NVL: F16. 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 phi3.5:3.8b-mini-instruct-fp16

Quantized files · GGUF

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

huggingface.co/bartowski/Phi-3.5-mini-instruct-GGUF (opens in a new tab)

ollama run hf.co/bartowski/Phi-3.5-mini-instruct-GGUF

Original weights · Hugging Face

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

huggingface.co/microsoft/Phi-3.5-mini-instruct (opens in a new tab)

No sign-in needed.

Official repo · GitHub

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

github.com/microsoft/PhiCookBook (opens in a new tab)

Microsoft’s official Phi cookbook: setup, inference and fine-tuning guides.

git clone https://github.com/microsoft/PhiCookBook

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 F16, whatever it runs on. Match score 61/100 (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 NVIDIA H100 NVL: 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.	4.0 / 15
Match score	62.0 / 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 NVIDIA H100 NVL

Quant	Memory needed	Available	Speed	Runs	Quality	Download
`Q2_K`	10.0 GB	94.0 GB	2,058 tok/s	Runs great	28Basic quality (D) · 28 of 100: how capable the model is at Q2_K, whatever it runs on. Match score 55/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q3_K_M`	10.4 GB	94.0 GB	1,620 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 57/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q4_K_M`	10.9 GB	94.0 GB	1,253 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 59/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q5_K_M`	11.2 GB	94.0 GB	1,035 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 60/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q6_K`	11.7 GB	94.0 GB	870 tok/s	Runs great	39Basic quality (D) · 39 of 100: how capable the model is at Q6_K, whatever it runs on. Match score 60/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`Q8_0`	12.6 GB	94.0 GB	638 tok/s	Runs great	40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 61/100 (quality, fit, speed and memory use) sets the ranking.	GGUF (opens in a new tab)
`F16`	16.2 GB	94.0 GB	294 tok/s	Runs great	40Fair quality (C) · 40 of 100: how capable the model is at F16, whatever it runs on. Match score 61/100 (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

40Fair quality (C) · 40 of 100: how capable the model is at F16, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. AMD Radeon RX 9070 XT F16 · 41.5 tok/s · Runs great
40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. Intel Arc B580 Q8_0 · 57.0 tok/s · Runs great
40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GeForce RTX 3060 12GB Q8_0 · 58.9 tok/s · Runs great
40Fair quality (C) · 40 of 100: how capable the model is at Q8_0, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GeForce RTX 4070 SUPER Q8_0 · 82.4 tok/s · Runs great
40Fair quality (C) · 40 of 100: how capable the model is at F16, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GeForce RTX 4070 Ti SUPER F16 · 50.7 tok/s · Runs great
40Fair quality (C) · 40 of 100: how capable the model is at F16, whatever it runs on. Match score 73/100 (quality, fit, speed and memory use) sets the ranking. NVIDIA GeForce RTX 4080 SUPER F16 · 55.5 tok/s · Runs great

Quick answers

How much memory does Phi-3.5 Mini Instruct need on the NVIDIA H100 NVL 94GB?

At F16 it needs about 16.2 GB. The NVIDIA H100 NVL 94GB has 94.0 GB usable.

How fast is Phi-3.5 Mini Instruct on the NVIDIA H100 NVL 94GB?

About 294 tok/s at F16, an estimate from the hardware’s memory bandwidth and the size of the model.

Which quantization of Phi-3.5 Mini Instruct should I use on the NVIDIA H100 NVL 94GB?

F16. 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.