Kokoro v0.19 TPS calculator

Open weights hexgrad 82M parameters December 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 · Q8_0 · 450 tok/s

Fastest card

B200

41,320 tok/s · 180 GB

Which GPUs can run Kokoro v0.19?

Set the inputs, read the answer

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

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

818 cards match

Calculating
Needs Quantisation Fit
41,320 tok/s

24,792–66,112 · low confidence

B200 NVIDIA 180 GB 8,000 GB/s Jan 2024 0.8 GB Q8_0 Comfortable
41,320 tok/s

24,792–66,112 · low confidence

B300 NVIDIA 288 GB 8,000 GB/s Sep 2025 0.8 GB Q8_0 Comfortable
32,995 tok/s

19,797–52,792 · low confidence

Radeon Instinct MI350X AMD 288 GB 8,190 GB/s Jan 2025 0.8 GB Q8_0 Comfortable
32,995 tok/s

19,797–52,792 · low confidence

Radeon Instinct MI355X AMD 288 GB 8,190 GB/s Jan 2025 0.8 GB Q8_0 Comfortable
26,388 tok/s

15,833–42,221 · low confidence

Radeon Instinct MI300 AMD 128 GB 6,550 GB/s Jan 2023 0.8 GB Q8_0 Comfortable
25,257 tok/s

15,154–40,411 · low confidence

H200 NVL NVIDIA 141 GB 4,890 GB/s Nov 2024 0.8 GB Q8_0 Comfortable
25,257 tok/s

15,154–40,411 · low confidence

H200 SXM 141 GB NVIDIA 141 GB 4,890 GB/s Nov 2024 0.8 GB Q8_0 Comfortable
24,172 tok/s

14,503–38,675 · low confidence

Radeon Instinct MI325X AMD 256 GB 6,000 GB/s Oct 2024 0.8 GB Q8_0 Comfortable
21,453 tok/s

12,872–34,324 · low confidence

Radeon Instinct MI300A AMD 128 GB 5,325 GB/s Dec 2023 0.8 GB Q8_0 Comfortable
21,453 tok/s

12,872–34,324 · low confidence

Radeon Instinct MI300X AMD 192 GB 5,325 GB/s Dec 2023 0.8 GB Q8_0 Comfortable
21,453 tok/s

12,872–34,324 · low confidence

Radeon Instinct MI308X AMD 192 GB 5,325 GB/s Dec 2023 0.8 GB Q8_0 Comfortable
20,350 tok/s

12,210–32,560 · low confidence

H100 NVL 94 GB NVIDIA 94 GB 3,940 GB/s Mar 2023 0.8 GB Q8_0 Comfortable
17,354 tok/s

10,413–27,767 · low confidence

H100 PCIe 96 GB NVIDIA 96 GB 3,360 GB/s Mar 2023 0.8 GB Q8_0 Comfortable
17,354 tok/s

10,413–27,767 · low confidence

H100 SXM5 80 GB NVIDIA 80 GB 3,360 GB/s Oct 2022 0.8 GB Q8_0 Comfortable
17,354 tok/s

10,413–27,767 · low confidence

H100 SXM5 94 GB NVIDIA 94 GB 3,360 GB/s Mar 2023 0.8 GB Q8_0 Comfortable
17,354 tok/s

10,413–27,767 · low confidence

H100 SXM5 96 GB NVIDIA 96 GB 3,360 GB/s Mar 2023 0.8 GB Q8_0 Comfortable
17,354 tok/s

10,413–27,767 · low confidence

H800 SXM5 NVIDIA 80 GB 3,360 GB/s Mar 2023 0.8 GB Q8_0 Comfortable
13,214 tok/s

7,928–21,143 · low confidence

Radeon Instinct MI250 AMD 128 GB 3,280 GB/s Nov 2021 0.8 GB Q8_0 Comfortable
13,214 tok/s

7,928–21,143 · low confidence

Radeon Instinct MI250X AMD 128 GB 3,280 GB/s Nov 2021 0.8 GB Q8_0 Comfortable
11,012 tok/s

6,607–17,619 · low confidence

Data Center GPU Max 1550 Intel 128 GB 3,280 GB/s Jan 2023 0.8 GB Q8_0 Comfortable
10,777 tok/s

6,466–17,243 · low confidence

Data Center GPU Max Subsystem Intel 128 GB 3,210 GB/s Jan 2023 0.8 GB Q8_0 Comfortable
10,537 tok/s

6,322–16,859 · low confidence

A100 SXM4 80 GB NVIDIA 80 GB 2,040 GB/s Nov 2020 0.8 GB Q8_0 Comfortable
10,537 tok/s

6,322–16,859 · low confidence

A100X NVIDIA 80 GB 2,040 GB/s Jun 2021 0.8 GB Q8_0 Comfortable
10,537 tok/s

6,322–16,859 · low confidence

A800 SXM4 80 GB NVIDIA 80 GB 2,040 GB/s Aug 2022 0.8 GB Q8_0 Comfortable
10,537 tok/s

6,322–16,859 · low confidence

H100 CNX NVIDIA 80 GB 2,040 GB/s Mar 2023 0.8 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
hexgrad
Published
25 December 2024

What it does

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

Domain
Speech
Task
Text-to-speech (TTS), Speech synthesis

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
82M

82M

Training data
tokens

"<100 hrs" of training audio data

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
1.7 × 10²⁰ FLOP

312000000000000 FLOP / GPU / sec * 500 GPU - hours * 3600 sec / hour * 0.3 [assumed utilization] = 1.6848e+20 FLOP

How it was established
Hardware

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 A100 SXM4 80 GB
Chip-hours
500

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

Apache 2.0 for weights https://huggingface.co/hexgrad/Kokoro-82M Apache 2.0 for inference code https://github.com/hexgrad/kokoro

Hugging Face
hexgrad

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
Kokoro v0.19
Last updated
28 November 2025

The extremes

What the numbers mean

Hardware requirements in practice

Minimum card

Tesla C1080

Memory needed

0.8 GB

Fastest

41,320 tok/s

Kokoro v0.19 is small enough at 82M 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, Q8_0 compression, roughly 450 tokens per second.

A B200 is the fastest we calculate for it: about 41,320 tokens per second, from 8,000 GB/s of memory bandwidth.

What this model is

Kokoro v0.19 was published by hexgrad, in December 2024.

It works in Speech, and is recorded as doing text-to-speech (TTS), Speech synthesis.

The weights being open is what puts this page in the calculator rather than only in the catalogue: it is a model you can actually hold. It is published under the hexgrad organisation on Hugging Face.

What decides the speed

The median result is around 1,160.3 tokens per second; 818 cards produce text faster than most people read it.

It is a dense model, so every parameter is read for every token produced. That makes speed track memory bandwidth almost exactly — a card with twice the bandwidth generates roughly twice as fast.

Without the attention layout on record, the memory column is an approximation. It is close enough to choose hardware by, and least reliable at long context.

What went into building it

Producing it required around 1.7 × 10²⁰ FLOP of arithmetic, on NVIDIA A100 SXM4 80 GB, which is a statement about the training budget rather than about inference.

Step by step

How to choose a GPU for Kokoro v0.19

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

  1. 01

    Start from the memory column

    The table lists every card that can hold Kokoro v0.19 — around 0.8 GB at Q8_0. That figure, not the card's headline performance, is what decides whether it runs.

  2. 02

    Set the context length you will work at

    The conversation occupies memory too, and grows as it goes. Set the slider to the length you expect: at long context Kokoro v0.19 can slip off a card that handles short questions easily.

  3. 03

    Choose how far you will compress it

    The quantisation column varies by card, because a bigger card holds a more accurate copy of Kokoro v0.19 — Q8_0 on the smallest card that fits. Set a floor to hold the comparison at one level.

  4. 04

    Sort by speed

    The speed ordering for Kokoro v0.19 is effectively an ordering by memory bandwidth, which is why the B200 tops it at 41,320 tok/s.

  5. 05

    Check the fit verdict before buying

    Tight means Kokoro v0.19 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.

  6. 06

    Check the card from the other side

    Each card page repeats this sweep for every model we hold. It answers what else the hardware is good for, beyond Kokoro v0.19.

Answers

Kokoro v0.19 — common questions

01

How much compute was used to train Kokoro v0.19?

Around 1.7 × 10²⁰ FLOP, on NVIDIA A100 SXM4 80 GB. 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.

02

Can I run Kokoro v0.19 if it does not fit in my GPU?

Only by offloading, which is usually a false economy: the part in system memory drags the whole thing down. Our figures for Kokoro v0.19 assume it is fully resident.

03

Would two GPUs run Kokoro v0.19 faster?

Capacity adds across cards; throughput does not. Since 818 of the cards we track already hold Kokoro v0.19 on their own, a second card is rarely the answer here.

04

Why does the quantisation differ between cards for Kokoro v0.19?

A larger card holds a more accurate copy. Across the cards that run Kokoro v0.19, 1 compression levels are used; the floor control above pins it to one.

05

How accurate are these Kokoro v0.19 speed estimates?

Every figure is derived from memory bandwidth and model size, not benchmarked. That is why each is published as a range such as 24,792–66,112 tok/s on the B200 rather than a single number.

06

What GPU do I need to run Kokoro v0.19?

The smallest card in our catalogue that holds Kokoro v0.19 is the Tesla C1080, with 4 GB of memory. It runs the model at Q8_0 using about 0.8 GB, and produces roughly 450 tokens per second. 818 cards in total can run it.

07

How fast is Kokoro v0.19 on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 41,320 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 818 of the cards that can run Kokoro v0.19 clear that.

08

How much VRAM does Kokoro v0.19 need?

About 0.8 GB at Q8_0 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.

09

Can I run Kokoro v0.19 on a 8 GB GPU?

Yes. A CMP 170HX 8 GB with 8 GB runs it at Q8_0, using about 0.8 GB and generating roughly 7,696 tokens per second — a comfortable fit.

10

Can I run Kokoro v0.19 on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q8_0, using about 0.8 GB and generating roughly 4,713 tokens per second — a comfortable fit.

11

Can I run Kokoro v0.19 on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q8_0, using about 0.8 GB and generating roughly 5,836 tokens per second — a comfortable fit.

12

Can I run Kokoro v0.19 on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 0.8 GB and generating roughly 6,921 tokens per second — a comfortable fit.

13

Is Kokoro v0.19 open source?

Its weights are published, so Kokoro v0.19 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.

14

How many parameters does Kokoro v0.19 have?

Kokoro v0.19 has 82M parameters. 82M. 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.

15

Who created Kokoro v0.19?

Kokoro v0.19 was published by hexgrad.

16

When was Kokoro v0.19 released?

Kokoro v0.19 was published in December 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.

17

What is Kokoro v0.19 used for?

Kokoro v0.19 works in Speech, and is recorded as handling text-to-speech (TTS), Speech synthesis. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

18

Where can I download Kokoro v0.19?

Its weights are published under the hexgrad 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.