Reason-ModernColBERT TPS calculator

Open weights LightOn 150M parameters May 2025

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 · 246 tok/s

Fastest card

B200

22,588 tok/s · 180 GB

Which GPUs can run Reason-ModernColBERT?

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
22,588 tok/s 13,553–36,141 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	0.9 GB	Q8_0	Comfortable
22,588 tok/s 13,553–36,141 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	0.9 GB	Q8_0	Comfortable
18,037 tok/s 10,822–28,860 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	0.9 GB	Q8_0	Comfortable
18,037 tok/s 10,822–28,860 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	0.9 GB	Q8_0	Comfortable
14,425 tok/s 8,655–23,081 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	0.9 GB	Q8_0	Comfortable
13,807 tok/s 8,284–22,091 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.9 GB	Q8_0	Comfortable
13,807 tok/s 8,284–22,091 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	0.9 GB	Q8_0	Comfortable
13,214 tok/s 7,928–21,143 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	0.9 GB	Q8_0	Comfortable
11,728 tok/s 7,037–18,764 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	0.9 GB	Q8_0	Comfortable
11,728 tok/s 7,037–18,764 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	0.9 GB	Q8_0	Comfortable
11,728 tok/s 7,037–18,764 · low confidence	Radeon Instinct MI308X AMD	192 GB	5,325 GB/s	Dec 2023	0.9 GB	Q8_0	Comfortable
11,125 tok/s 6,675–17,800 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	0.9 GB	Q8_0	Comfortable
9,487 tok/s 5,692–15,179 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.9 GB	Q8_0	Comfortable
9,487 tok/s 5,692–15,179 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	0.9 GB	Q8_0	Comfortable
9,487 tok/s 5,692–15,179 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	0.9 GB	Q8_0	Comfortable
9,487 tok/s 5,692–15,179 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	0.9 GB	Q8_0	Comfortable
9,487 tok/s 5,692–15,179 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	0.9 GB	Q8_0	Comfortable
7,224 tok/s 4,334–11,558 · low confidence	Radeon Instinct MI250 AMD	128 GB	3,280 GB/s	Nov 2021	0.9 GB	Q8_0	Comfortable
7,224 tok/s 4,334–11,558 · low confidence	Radeon Instinct MI250X AMD	128 GB	3,280 GB/s	Nov 2021	0.9 GB	Q8_0	Comfortable
6,020 tok/s 3,612–9,632 · low confidence	Data Center GPU Max 1550 Intel	128 GB	3,280 GB/s	Jan 2023	0.9 GB	Q8_0	Comfortable
5,891 tok/s 3,535–9,426 · low confidence	Data Center GPU Max Subsystem Intel	128 GB	3,210 GB/s	Jan 2023	0.9 GB	Q8_0	Comfortable
5,760 tok/s 3,456–9,216 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	0.9 GB	Q8_0	Comfortable
5,760 tok/s 3,456–9,216 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	0.9 GB	Q8_0	Comfortable
5,760 tok/s 3,456–9,216 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	0.9 GB	Q8_0	Comfortable
5,760 tok/s 3,456–9,216 · low confidence	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	0.9 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: LightOn
Organisation type: Industry
Country: France
Published: 22 May 2025
Authors: Antoine Chaffin

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, Quantitative reasoning, Retrieval-augmented generation
Base model: GTE-ModernColBERT-v1

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: 150M
Training data: tokens
Epochs: 3

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 × 10²¹ FLOP
How it was established: Operation counting
Fine-tuning compute: 9.6 × 10¹⁶ 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: 2 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 (non-commercial)
Training code: Open source
Hugging Face: lightonai

How it is classified

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

Record confidence: Likely

Sources

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

Reference: LightOn Unlocks Agentic RAG with new SOTA Model Reason-ModernColBERT
Last updated: 28 November 2025

The extremes

The ten fastest GPUs that run Reason-ModernColBERT

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 Reason-ModernColBERT

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

0.9 GB

Fastest

22,588 tok/s

Reason-ModernColBERT is small enough at 150M 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 246 tokens per second.

A B200 is the fastest we calculate for it: about 22,588 tokens per second, from 8,000 GB/s of memory bandwidth.

What this model is

Reason-ModernColBERT was published by LightOn, in France, in May 2025. industry is the category the publisher falls under.

It works in Language, and is recorded as doing language modeling/generation, Question answering, Quantitative reasoning, Retrieval-augmented generation.

Its starting point was GTE-ModernColBERT-v1 — most models at this scale are adapted from an existing base rather than built from nothing.

Published weights mean the model runs on your machine rather than someone else's, which is what makes the hardware question below answerable at all. It is published under the lightonai organisation on Hugging Face.

What decides the speed

Across every card that can run it, the middle of the range is about 634.3 tokens per second, and 818 of them clear the ten tokens per second that roughly matches reading speed.

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

The training run consumed about 3.7 × 10²¹ FLOP. That figure describes the cost of creating it and has no bearing on how quickly it generates text.

Step by step

How to choose a GPU for Reason-ModernColBERT

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

01

Start from the memory column

The table lists every card that can hold Reason-ModernColBERT — around 0.9 GB at Q8_0. That figure, not the card's headline performance, is what decides whether it runs.
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 Reason-ModernColBERT 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 — Q8_0 on the smallest card that fits. Setting a floor drops the cards that only manage Reason-ModernColBERT by squeezing it further than you would want.
04

Rank by throughput rather than spec sheet

Sort by speed to see how cards rank for Reason-ModernColBERT. It will not match a gaming ordering — generation is bound by memory bandwidth, which is why the B200 tops it at 22,588 tok/s.
05

Check the fit verdict before buying

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

Answers

Reason-ModernColBERT — common questions

What is Reason-ModernColBERT used for?

Reason-ModernColBERT works in Language, and is recorded as handling language modeling/generation, Question answering, Quantitative reasoning, Retrieval-augmented generation. These are the areas it was designed around; they describe intent rather than a hard boundary.

Where can I download Reason-ModernColBERT?

Its weights are published under the lightonai 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 Reason-ModernColBERT?

Around 3.7 × 10²¹ FLOP. 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 Reason-ModernColBERT if it does not fit in my GPU?

It can be split between the card and system memory, but Reason-ModernColBERT generates painfully slowly that way. Nothing on this page assumes offloading.

Would two GPUs run Reason-ModernColBERT faster?

A second card roughly doubles the memory available but not the generation rate. With 818 cards already able to run Reason-ModernColBERT alone, the case for pairing is weak.

Why does the quantisation differ between cards for Reason-ModernColBERT?

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

How accurate are these Reason-ModernColBERT speed estimates?

They are calculated from specifications rather than measured, and each carries a range — 13,553–36,141 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 Reason-ModernColBERT?

The smallest card in our catalogue that holds Reason-ModernColBERT is the Tesla C1080, with 4 GB of memory. It runs the model at Q8_0 using about 0.9 GB, and produces roughly 246 tokens per second. 818 cards in total can run it.

How fast is Reason-ModernColBERT on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 22,588 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 Reason-ModernColBERT clear that.

How much VRAM does Reason-ModernColBERT need?

About 0.9 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.

Can I run Reason-ModernColBERT on a 8 GB GPU?

Yes. A CMP 170HX 8 GB with 8 GB runs it at Q8_0, using about 0.9 GB and generating roughly 4,207 tokens per second — a comfortable fit.

Can I run Reason-ModernColBERT on a 12 GB GPU?

Yes. A GeForce RTX 3080 Ti with 12 GB runs it at Q8_0, using about 0.9 GB and generating roughly 2,576 tokens per second — a comfortable fit.

Can I run Reason-ModernColBERT on a 16 GB GPU?

Yes. A Tesla V100 SXM2 16 GB with 16 GB runs it at Q8_0, using about 0.9 GB and generating roughly 3,191 tokens per second — a comfortable fit.

Can I run Reason-ModernColBERT on a 24 GB GPU?

Yes. A GeForce RTX 5090 D V2 with 24 GB runs it at Q8_0, using about 0.9 GB and generating roughly 3,784 tokens per second — a comfortable fit.

Is Reason-ModernColBERT open source?

Its weights are published, so Reason-ModernColBERT 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 Reason-ModernColBERT have?

Reason-ModernColBERT has 150M parameters. 150M. 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 Reason-ModernColBERT?

Reason-ModernColBERT was published by LightOn, based in France, categorised as industry.

When was Reason-ModernColBERT released?

Reason-ModernColBERT was published in May 2025.

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