Unicorn TPS calculator

Open weights Allen Institute for AI 11B parameters March 2021

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

509 cards that can run it

818 cards we hold specifications for

Smallest card that fits

Xeon Phi 5110P

8 GB · IQ4_XS · 19.7 tok/s

Fastest card

B200

308 tok/s · 180 GB

Which GPUs can run Unicorn?

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

509 cards match

Calculating

					Needs	Quantisation	Fit
308 tok/s 185–493 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	12.5 GB	Q8_0	Comfortable
308 tok/s 185–493 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	12.5 GB	Q8_0	Comfortable
246 tok/s 148–394 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	12.5 GB	Q8_0	Comfortable
246 tok/s 148–394 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	12.5 GB	Q8_0	Comfortable
197 tok/s 118–315 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	12.5 GB	Q8_0	Comfortable
188 tok/s 113–301 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	12.5 GB	Q8_0	Comfortable
188 tok/s 113–301 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	12.5 GB	Q8_0	Comfortable
180 tok/s 108–288 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	12.5 GB	Q8_0	Comfortable
160 tok/s 96–256 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	12.5 GB	Q8_0	Comfortable
160 tok/s 96–256 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	12.5 GB	Q8_0	Comfortable
160 tok/s 96–256 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	12.5 GB	Q8_0	Comfortable
152 tok/s 91–243 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	12.5 GB	Q8_0	Comfortable
141 tok/s 85–225 · low confidence	CMP 170HX 8 GB NVIDIA	8 GB	1,490 GB/s	Sep 2021	6.7 GB	IQ4_XS	Tight
129 tok/s 78–207 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	12.5 GB	Q8_0	Comfortable
129 tok/s 78–207 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	12.5 GB	Q8_0	Comfortable
129 tok/s 78–207 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	12.5 GB	Q8_0	Comfortable
129 tok/s 78–207 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	12.5 GB	Q8_0	Comfortable
129 tok/s 78–207 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	12.5 GB	Q8_0	Comfortable
107 tok/s 64–172 · low confidence	CMP 170HX 10 GB NVIDIA	10 GB	1,560 GB/s	Sep 2021	8.6 GB	Q5_K_M	Tight
98.5 tok/s 59–158 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	12.5 GB	Q8_0	Comfortable
98.5 tok/s 59–158 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	12.5 GB	Q8_0	Comfortable
82.1 tok/s 49–131 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	12.5 GB	Q8_0	Comfortable
80.3 tok/s 48–129 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	12.5 GB	Q8_0	Comfortable
78.6 tok/s 47–126 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	12.5 GB	Q8_0	Comfortable
78.6 tok/s 47–126 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	12.5 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: Allen Institute for AI
Organisation type: Research collective
Country: United States of America
Published: 24 March 2021
Authors: Nicholas Lourie, Ronan Le Bras, Chandra Bhagavatula, Yejin Choi

What it does

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

Domain: Language
Task: Question answering, Language modeling/generation
Base model: T5-11B

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: 11B
Training data: tokens
Epochs: 2.46
Batch size: 32

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.

How it was established: Hardware
Fine-tuning compute: 3.3 × 10²² FLOP

The training run

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

Wall-clock time: 3 hours

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: Open source

How it is classified

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

Why it is tracked: SOTA improvement
Record confidence: Confident

Sources

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

Reference: UNICORN on RAINBOW: A Universal Commonsense Reasoning Model on a New Multitask Benchmark
Last updated: 28 November 2025

The extremes

The ten fastest GPUs that run Unicorn

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 Unicorn

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

What you need to run it

Minimum card

Xeon Phi 5110P

Memory needed

6.7 GB

Fastest

308 tok/s

Unicorn reaches a parameter count of 11B. That is small enough that hardware is rarely the obstacle, including on cards several years old. The number of cards we track that can run it: 509.

The least hardware that works is Xeon Phi 5110P, with a memory capacity of 8 GB, running it at a compression of IQ4_XS and producing around 19.7 tokens per second.

At the other end sits B200, generating roughly 308 tokens per second on the strength of a memory bandwidth of 8,000 GB/s.

Where it came from

Unicorn was published by Allen Institute for AI, in the country recorded as United States of America, during March 2021. The publishing organisation is categorised as research collective.

It works in the domain of Language, and is recorded as performing the task of question answering, Language modeling/generation.

Rather than being trained from scratch, it is derived from T5-11B. That is why it shares the base model's general shape and size.

The weights are published, so it can be downloaded and run on your own hardware indefinitely, offline, with no account attached.

Understanding the speeds

The median result is around 21.2 tokens per second. Clearing the ten tokens per second that roughly matches reading speed: 460 of them.

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.

Training and provenance

It is tracked in the underlying dataset for one reason in particular: sOTA improvement.

Step by step

How to choose a GPU for Unicorn

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

Start from what it actually needs, which is the requirement of Unicorn, needing around 6.7 GB at a compression of IQ4_XS. That figure, not the headline performance of a card, is what decides whether it runs.
02

Decide how long your conversations run

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

Choose how far you will compress it

Compression is what makes a model fit smaller cards, at some cost in accuracy, reaching a compression of IQ4_XS on the smallest card that fits. Setting a minimum quality drops the cards that only manage it by squeezing further than you would want, and holds the comparison at one level.
04

Compare tokens per second, not specifications

Ranking by tokens per second follows memory bandwidth rather than core counts, for Unicorn. It will not match a gaming ordering, because generation is bound by memory bandwidth. The card topping the list is B200, at 308 tok/s.
05

Check the fit verdict before buying

A tight fit runs, but leaves nothing spare for a longer conversation, in the case of Unicorn. Comfortable means you can grow the context later. That difference matters more than a few tokens per second, so buy for comfortable if you expect to.
06

Open the card you have settled on

Each card page repeats this sweep for every model we hold, answering what else the hardware is good for beyond Unicorn.

Answers

Unicorn — common questions

Unicorn— how fast is it on a GPU?

It depends on the card. The quickest we calculate is B200, at about 308 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and the number of cards clearing that: 460.

Unicorn— how much VRAM does it need?

It needs about 6.7 GB at a compression of IQ4_XS, 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.

Unicorn— can I run it on a GPU holding 8 GB?

Yes. The card CMP 170HX 8 GB, holding 8 GB, runs it at a compression of IQ4_XS, using about 6.7 GB and generating roughly 141 tokens per second. The fit is tight.

Unicorn— can I run it on a GPU holding 12 GB?

Yes. The card GeForce RTX 3080 Ti, holding 12 GB, runs it at a compression of Q6_K, using about 9.9 GB and generating roughly 51.0 tokens per second. The fit is tight.

Unicorn— can I run it on a GPU holding 16 GB?

Yes. The card Tesla V100 SXM2 16 GB, holding 16 GB, runs it at a compression of Q8_0, using about 12.5 GB and generating roughly 43.5 tokens per second. The fit is tight.

Unicorn— can I run it on a GPU holding 24 GB?

Yes. The card GeForce RTX 5090 D V2, holding 24 GB, runs it at a compression of Q8_0, using about 12.5 GB and generating roughly 51.6 tokens per second. The fit is comfortable.

Unicorn— is it open source?

Its weights are published, so it 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.

Unicorn— how many parameters does it have?

It has a parameter count of 11B. 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.

Unicorn— who created it?

It was published by Allen Institute for AI, based in United States of America, an organisation categorised as research collective.

Unicorn— when was it released?

It was published in March 2021. Capability per parameter has improved considerably since, so a newer model of the same size is often the better use of the same hardware.

Unicorn— what is it used for?

It works in the domain of Language, and is recorded as handling the task of question answering, Language modeling/generation. These are the areas it was designed around; they describe intent rather than a hard boundary.

Unicorn— where can I download it?

The weights are published, though we do not hold a repository link for it. This site calculates hardware requirements rather than hosting model files.

Unicorn— can I run it if it does not fit in my GPU?

It can be split between the card and system memory, but it generates painfully slowly that way. The nearest miss we calculate falls short by 2.0 GB. Every figure here assumes the whole model is resident on the card.

Unicorn— would two GPUs run it faster?

Two cards buy memory rather than speed, which matters only if one card cannot hold it. The number that can: 509. So a second card is rarely the answer here.

Unicorn— why does the quantisation differ between cards?

Each card is shown running the least-compressed copy it can hold. The number of distinct compression levels across the cards that fit it: 4. Bigger cards get the more accurate version, and the quality floor above pins the comparison to one level.

Unicorn— how accurate are these speed estimates?

They are calculated from specifications rather than measured, and each carries a range. One example: 185–493 tok/s on B200. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

Unicorn— what GPU do I need to run it?

The smallest card in our catalogue that holds it is Xeon Phi 5110P, with a memory capacity of 8 GB. It runs the model at a compression of IQ4_XS using about 6.7 GB, and produces roughly 19.7 tokens per second. The number of cards able to run it in total: 509.

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