phi-3.5-mini TPS calculator

Open weights Microsoft 3.8B parameters April 2024

Each card below is assessed against this model at the context length and minimum quality you choose. Speed is an estimate for a single request, calculated from the card's memory bandwidth and the size of the model once compressed.

Calculated for this model

818 cards that can run it

818 cards we hold specifications for

Smallest card that fits

Tesla C1080

4 GB · Q5_K_M · 17.3 tok/s

Fastest card

B200

892 tok/s · 180 GB

Which GPUs can run phi-3.5-mini?

Set the inputs, read the answer

Context length tokens

A longer conversation needs more memory, which can push this model off smaller cards.

Minimum quality

Hides cards that would only fit the model by compressing it below this point.

Search cards

Card memory

Also show cards that cannot run it

818 cards match

Calculating

					Needs	Quantisation	Fit
892 tok/s 535–1,427 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	4.8 GB	Q8_0	Comfortable
892 tok/s 535–1,427 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	4.8 GB	Q8_0	Comfortable
712 tok/s 427–1,139 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	4.8 GB	Q8_0	Comfortable
712 tok/s 427–1,139 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	4.8 GB	Q8_0	Comfortable
569 tok/s 342–911 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	4.8 GB	Q8_0	Comfortable
545 tok/s 327–872 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	4.8 GB	Q8_0	Comfortable
545 tok/s 327–872 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	4.8 GB	Q8_0	Comfortable
522 tok/s 313–835 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	4.8 GB	Q8_0	Comfortable
463 tok/s 278–741 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	4.8 GB	Q8_0	Comfortable
463 tok/s 278–741 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	4.8 GB	Q8_0	Comfortable
463 tok/s 278–741 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	4.8 GB	Q8_0	Comfortable
439 tok/s 263–703 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	4.8 GB	Q8_0	Comfortable
374 tok/s 225–599 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	4.8 GB	Q8_0	Comfortable
374 tok/s 225–599 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	4.8 GB	Q8_0	Comfortable
374 tok/s 225–599 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	4.8 GB	Q8_0	Comfortable
374 tok/s 225–599 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	4.8 GB	Q8_0	Comfortable
374 tok/s 225–599 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	4.8 GB	Q8_0	Comfortable
285 tok/s 171–456 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	4.8 GB	Q8_0	Comfortable
285 tok/s 171–456 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	4.8 GB	Q8_0	Comfortable
238 tok/s 143–380 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	4.8 GB	Q8_0	Comfortable
233 tok/s 140–372 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	4.8 GB	Q8_0	Comfortable
227 tok/s 136–364 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	4.8 GB	Q8_0	Comfortable
227 tok/s 136–364 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	4.8 GB	Q8_0	Comfortable
227 tok/s 136–364 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	4.8 GB	Q8_0	Comfortable
227 tok/s 136–364 · low confidence	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	4.8 GB	Q8_0	Comfortable

Speeds are estimates for a single request — one conversation at a time — calculated from memory bandwidth, model size and quantisation. Real throughput varies with the inference runtime and its version. Figures published by hardware vendors measure many simultaneous requests and are much higher.

On record

Full specification

Everything on record for this model. Most of it describes how it was trained rather than how it runs — useful context for judging how much work went into it, and how it compares with models built at a different scale.

Origin

Who built this model, where, and when it was published.

Organisation: Microsoft
Organisation type: Industry
Country: United States of America
Published: 23 April 2024
Authors: Marah Abdin, Sam Ade Jacobs, Ammar Ahmad Awan, Jyoti Aneja, Ahmed Awadallah, Hany Awadalla, Nguyen Bach, Amit Bahree, Arash Bakhtiari, Harkirat Behl, Alon Benhaim, Misha Bilenko, Johan Bjorck, Sébastien Bubeck, Martin Cai, Caio César Teodoro Mendes, Weizhu Chen, Vishrav Chaudhary, Parul Chopra, Allie Del Giorno, Gustavo de Rosa, Matthew Dixon, Ronen Eldan, Dan Iter, Amit Garg, Abhishek Goswami, Su…

What it does

The problem areas the model was built for. A model can carry several of each.

Domain: Language
Task: Language modeling/generation, Question answering, Translation

Size

How large the model is and how much data it was trained on. Parameters are the figure that decides whether it fits on a given graphics card.

Parameters: 3.8B
Training data: tokens

Training compute

The arithmetic performed to train the model, measured in floating-point operations. It is a measure of what the training run cost, not of how fast the finished model answers you.

Training compute: 3.7 × 10²² FLOP
How it was established: Hardware,Operation counting

The training run

What it physically took to train: which chips, how many, for how long, and what that drew from the wall.

Training hardware: NVIDIA H100 SXM5 80GB
Chips used: 512
Wall-clock time: 240 hours (10 days)
Power draw: 708.4 kW

Availability

Whether you can obtain the model and run it on your own hardware, which is what decides if any of the graphics-card figures on this page apply.

Weights: Open — downloadable
Model access: Open weights (unrestricted)
Training code: Unreleased
Hugging Face: microsoft

How it is classified

Labels the source dataset applies when tracking notable models, and how confident it is in the entry.

Record confidence: Confident

Sources

Where this record came from and when it was last checked.

Reference: Phi-3 Technical Report: A Highly Capable Language Model Locally on Your Phone
Last updated: 28 November 2025

The extremes

The ten fastest GPUs that run phi-3.5-mini

Ranked by estimated tokens per second, newest card first where speeds tie. Because generation is bound by memory bandwidth, this ordering follows bandwidth rather than any gaming benchmark.

The smallest GPUs that still run phi-3.5-mini

The cheapest route in, by memory capacity. A tight fit runs the model but leaves nothing spare for a longer conversation.

What the numbers mean

Hardware requirements in practice

Minimum card

Tesla C1080

Memory needed

3.4 GB

Fastest

892 tok/s

phi-3.5-mini is small enough at 3.8B parameters that hardware is rarely the obstacle — 818 of the cards we track can run it, including cards several years old.

The entry point is the Tesla C1080: 4 GB of memory, Q5_K_M compression, roughly 17.3 tokens per second.

At the other end, a B200 generates roughly 892 tokens per second on it, on the strength of 8,000 GB/s of memory bandwidth.

What this model is

phi-3.5-mini was published by Microsoft, in United States of America, in April 2024. The organisation is categorised as industry.

It works in Language, and is recorded as doing language modeling/generation, Question answering, Translation.

Because its weights were released, nothing about running it depends on a provider staying available — it is yours once downloaded. It is published under the microsoft organisation on Hugging Face.

What decides the speed

Across every card that can run it, the middle of the range is about 29.9 tokens per second, and 777 of them clear the ten tokens per second that roughly matches reading speed.

Being dense, it reads all of itself per token, which is why the ordering by speed below follows the ordering by memory bandwidth so closely.

Memory here is estimated from size rather than computed from the architecture, which is not recorded for this model — the numbers are indicative rather than exact.

How it was trained

Producing it required around 3.7 × 10²² FLOP of arithmetic, on NVIDIA H100 SXM5 80GB, which is a statement about the training budget rather than about inference.

Step by step

How to choose a GPU for phi-3.5-mini

The table above has already assessed every card we hold specifications for against this model. Getting to your answer takes six steps.

01

Read the memory figure first

The table lists every card that can hold phi-3.5-mini — around 3.4 GB at Q5_K_M. That figure, not the card's headline performance, is what decides whether it runs.
02

Set the context length you will work at

Longer conversations cost memory on top of what the weights need. Move the slider to your real working length before trusting any row for phi-3.5-mini.
03

Choose how far you will compress it

The quantisation column varies by card, because a bigger card holds a more accurate copy of phi-3.5-mini — Q5_K_M on the smallest card that fits. Set a floor to hold the comparison at one level.
04

Compare tokens per second, not specifications

The speed ordering for phi-3.5-mini is effectively an ordering by memory bandwidth, which is why the B200 tops it at 892 tok/s.
05

Look at the headroom, not just the fit

Tight means phi-3.5-mini loads and works, with no room to raise the context later. Comfortable means you can. The difference matters more than a few tokens per second.
06

Check the card from the other side

Following a card through to its own page shows every other model it can hold, which is the question that follows once phi-3.5-mini is settled.

Answers

phi-3.5-mini — common questions

How much compute was used to train phi-3.5-mini?

Around 3.7 × 10²² FLOP, on NVIDIA H100 SXM5 80GB. That measures what producing the model cost and says nothing about how quickly it answers once trained — inference speed comes from memory bandwidth, not from the training budget.

Can I run phi-3.5-mini if it does not fit in my GPU?

Partly. Layers that do not fit sit in system memory and run at a fraction of the speed, so a mostly-offloaded phi-3.5-mini is rarely worth using. Every figure here assumes the whole model is on the card.

Would two GPUs run phi-3.5-mini faster?

Two cards buy memory rather than speed. That matters for phi-3.5-mini only if one card cannot hold it — 818 can, so a second adds little.

Why does the quantisation differ between cards for phi-3.5-mini?

Each card is shown running the least-compressed copy it can hold, and phi-3.5-mini appears at 3 different compression levels across the cards that fit it. Bigger cards get the more accurate version.

How accurate are these phi-3.5-mini speed estimates?

They are calculated from specifications rather than measured, and each carries a range — 535–1,427 tok/s on the B200, for instance. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

What GPU do I need to run phi-3.5-mini?

The smallest card in our catalogue that holds phi-3.5-mini is the Tesla C1080, with 4 GB of memory. It runs the model at Q5_K_M using about 3.4 GB, and produces roughly 17.3 tokens per second. 818 cards in total can run it.

How fast is phi-3.5-mini on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 892 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 777 of the cards that can run phi-3.5-mini clear that.

How much VRAM does phi-3.5-mini need?

About 3.4 GB at Q5_K_M compression, which is what the smallest card that runs it uses. Less compression needs more: the figures in the memory column above are recalculated for each card, because each one holds the least-compressed version it can.

Can I run phi-3.5-mini on a 8 GB GPU?

Yes. A CMP 170HX 8 GB with 8 GB runs it at Q8_0, using about 4.8 GB and generating roughly 166 tokens per second — a comfortable fit.

Can I run phi-3.5-mini on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q8_0, using about 4.8 GB and generating roughly 102 tokens per second — a comfortable fit.

Can I run phi-3.5-mini on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q8_0, using about 4.8 GB and generating roughly 126 tokens per second — a comfortable fit.

Can I run phi-3.5-mini on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 4.8 GB and generating roughly 149 tokens per second — a comfortable fit.

Is phi-3.5-mini open source?

Its weights are published, so phi-3.5-mini can be downloaded and run on your own hardware. Note that open weights is not the same as open source in the full sense — it says nothing about the training data, the training code, or the commercial terms attached.

How many parameters does phi-3.5-mini have?

phi-3.5-mini has 3.8B parameters. 3.8B. That figure is the total, and it is what decides how much memory the model needs — roughly half a gigabyte per billion at the compression most people use.

Who created phi-3.5-mini?

phi-3.5-mini was published by Microsoft, based in United States of America, categorised as industry.

When was phi-3.5-mini released?

phi-3.5-mini was published in April 2024. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

What is phi-3.5-mini used for?

phi-3.5-mini works in Language, and is recorded as handling language modeling/generation, Question answering, Translation. These are the areas it was designed around; they describe intent rather than a hard boundary.

Where can I download phi-3.5-mini?

Its weights are published under the microsoft organisation on Hugging Face. We do not host model files — this site calculates what hardware is needed to run them.

Source

Original publication

Record last updated 28 November 2025

The other direction

Looking at it from the other side?

This page starts from the model. If you already own a card and want to know everything it will run, start from the hardware instead.

All AI models