nekomata-14b TPS calculator

Open weights rinna 14.2B parameters December 2023

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

306 cards that can run it

818 cards we hold specifications for

Smallest card that fits

P102-101

10 GB · IQ4_XS · 19.9 tok/s

Fastest card

B200

239 tok/s · 180 GB

Which GPUs can run nekomata-14b?

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

306 cards match

Calculating

					Needs	Quantisation	Fit
239 tok/s 143–382 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	15.9 GB	Q8_0	Comfortable
239 tok/s 143–382 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	15.9 GB	Q8_0	Comfortable
191 tok/s 114–305 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	15.9 GB	Q8_0	Comfortable
191 tok/s 114–305 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	15.9 GB	Q8_0	Comfortable
152 tok/s 91–244 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	15.9 GB	Q8_0	Comfortable
146 tok/s 88–233 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	15.9 GB	Q8_0	Comfortable
146 tok/s 88–233 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	15.9 GB	Q8_0	Comfortable
140 tok/s 84–223 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	15.9 GB	Q8_0	Comfortable
124 tok/s 74–198 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	15.9 GB	Q8_0	Comfortable
124 tok/s 74–198 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	15.9 GB	Q8_0	Comfortable
124 tok/s 74–198 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	15.9 GB	Q8_0	Comfortable
118 tok/s 71–188 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	15.9 GB	Q8_0	Comfortable
114 tok/s 69–183 · low confidence	CMP 170HX 10 GB NVIDIA	10 GB	1,560 GB/s	Sep 2021	8.5 GB	IQ4_XS	Tight
100 tok/s 60–160 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	15.9 GB	Q8_0	Comfortable
100 tok/s 60–160 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	15.9 GB	Q8_0	Comfortable
100 tok/s 60–160 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	15.9 GB	Q8_0	Comfortable
100 tok/s 60–160 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	15.9 GB	Q8_0	Comfortable
100 tok/s 60–160 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	15.9 GB	Q8_0	Comfortable
76.3 tok/s 46–122 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	15.9 GB	Q8_0	Comfortable
76.3 tok/s 46–122 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	15.9 GB	Q8_0	Comfortable
63.6 tok/s 38–102 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	15.9 GB	Q8_0	Comfortable
62.8 tok/s 38–101 · low confidence	GeForce RTX 3080 12 GB NVIDIA	12 GB	912 GB/s	Jan 2022	9.3 GB	Q4_K_M	Tight
62.8 tok/s 38–101 · low confidence	GeForce RTX 3080 Ti NVIDIA	12 GB	912 GB/s	May 2021	9.3 GB	Q4_K_M	Tight
62.2 tok/s 37–100 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	15.9 GB	Q8_0	Comfortable
60.9 tok/s 37–97 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	15.9 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: rinna
Organisation type: Industry
Country: Japan
Published: 21 December 2023
Authors: Tianyu Zhao, Akio Kaga, Kei Sawada

What it does

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

Domain: Language
Task: Language generation
Base model: Qwen-14B
Numerical format: BF16

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: 14.2B
Training data: 66,000,000,000 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: 2.6 × 10²³ FLOP
How it was established: 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: Amazon Trainium1
Chips used: 256
Wall-clock time: 168 hours (7 days)
Hardware utilisation: MFU 17.1%

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 (restricted use)
Training code: Unreleased
Hugging Face: rinna

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
Citations: 35

Sources

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

Reference: rinna/nekomata-14b
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run nekomata-14b

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 nekomata-14b

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

P102-101

Memory needed

8.5 GB

Fastest

239 tok/s

nekomata-14b is small enough at 14.2B parameters that hardware is rarely the obstacle — 306 of the cards we track can run it, including cards several years old.

The entry point is the P102-101: 10 GB of memory, IQ4_XS compression, roughly 19.9 tokens per second.

Top of the range is the B200, at roughly 239 tokens per second thanks to 8,000 GB/s of bandwidth.

Background

nekomata-14b was published by rinna, in Japan, in December 2023. The organisation is categorised as industry.

It works in Language, and is recorded as doing language generation.

It builds on Qwen-14B, which is why it shares that 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. It is published under the rinna organisation on Hugging Face.

Reading the throughput figures

The median result is around 22.3 tokens per second; 266 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.

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

Producing it required around 2.6 × 10²³ FLOP of arithmetic, on Amazon Trainium1, which is a statement about the training budget rather than about inference.

Around 66,000,000,000 tokens went into training it.

The reason it appears in this catalogue at all is sOTA improvement.

Step by step

How to choose a GPU for nekomata-14b

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

01

Check what it needs before anything else

Look at what nekomata-14b actually needs — around 8.5 GB at IQ4_XS. No amount of processing power compensates for a card that cannot hold it.
02

Match the context to your actual use

The conversation occupies memory too, and grows as it goes. Set the slider to the length you expect: at long context nekomata-14b can slip off a card that handles short questions easily.
03

Choose how far you will compress it

Each card runs the least-compressed copy it can hold — IQ4_XS on the smallest card that fits. Setting a floor drops the cards that only manage nekomata-14b by squeezing it further than you would want.
04

Sort by speed

Ranking by tokens per second for nekomata-14b follows memory bandwidth, not core counts, which is why the B200 tops it at 239 tok/s.
05

Check the fit verdict before buying

The fit column separates cards that just manage nekomata-14b from those with room to spare. Buy for the second if the context might grow.
06

Open the card you have settled on

Every card name links to its own page, which runs the same calculation across the whole model catalogue. Worth a look before buying for nekomata-14b alone — a card is usually bought for more than one model.

Answers

nekomata-14b — common questions

Would two GPUs run nekomata-14b faster?

Two cards buy memory rather than speed. That matters for nekomata-14b only if one card cannot hold it — 306 can, so a second adds little.

Why does the quantisation differ between cards for nekomata-14b?

Each card is shown running the least-compressed copy it can hold, and nekomata-14b appears at 4 different compression levels across the cards that fit it. Bigger cards get the more accurate version.

How accurate are these nekomata-14b 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 143–382 tok/s on the B200 rather than a single number.

What GPU do I need to run nekomata-14b?

The smallest card in our catalogue that holds nekomata-14b is the P102-101, with 10 GB of memory. It runs the model at IQ4_XS using about 8.5 GB, and produces roughly 19.9 tokens per second. 306 cards in total can run it.

How fast is nekomata-14b on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 239 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 266 of the cards that can run nekomata-14b clear that.

How much VRAM does nekomata-14b need?

About 8.5 GB at IQ4_XS 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 nekomata-14b on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q4_K_M, using about 9.3 GB and generating roughly 62.8 tokens per second — a tight fit.

Can I run nekomata-14b on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q6_K, using about 12.6 GB and generating roughly 49.0 tokens per second — a tight fit.

Can I run nekomata-14b on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 15.9 GB and generating roughly 40.0 tokens per second — a comfortable fit.

Is nekomata-14b open source?

Its weights are published, so nekomata-14b 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 nekomata-14b have?

nekomata-14b has 14.2B parameters. Source: https://huggingface.co/rinna/nekomata-14b. 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 nekomata-14b?

nekomata-14b was published by rinna, based in Japan, categorised as industry.

When was nekomata-14b released?

nekomata-14b was published in December 2023. 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 nekomata-14b used for?

nekomata-14b works in Language, and is recorded as handling language generation. These are the areas it was designed around; they describe intent rather than a hard boundary.

Where can I download nekomata-14b?

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

How much compute was used to train nekomata-14b?

Around 2.6 × 10²³ FLOP, on Amazon Trainium1. 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 nekomata-14b 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 — the nearest miss we calculate is short by 2.1 GB. Our figures for nekomata-14b assume it is fully resident.

Source

Original publication

Record last updated 25 May 2026

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