FLM-101B TPS calculator

Open weights Chinese Academy of Sciences,Harbin Institute of Technology,Nanyang Technological University,Beijing Academy of Artificial Intelligence / BAAI 101B parameters September 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

43 cards that can run it

818 cards we hold specifications for

Smallest card that fits

Radeon Instinct MI200

64 GB · IQ4_XS · 13.2 tok/s

Fastest card

B200

33.6 tok/s · 180 GB

Which GPUs can run FLM-101B?

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

43 cards match

Calculating

					Needs	Quantisation	Fit
33.6 tok/s 20–54 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	108.8 GB	Q8_0	Comfortable
33.6 tok/s 20–54 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	108.8 GB	Q8_0	Comfortable
32.5 tok/s 20–52 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	61.8 GB	Q4_K_M	Tight
32.5 tok/s 20–52 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	61.8 GB	Q4_K_M	Tight
29.5 tok/s 18–47 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	73.6 GB	Q5_K_M	Tight
26.8 tok/s 16–43 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	108.8 GB	Q8_0	Comfortable
26.8 tok/s 16–43 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	108.8 GB	Q8_0	Comfortable
25.2 tok/s 15–40 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	73.6 GB	Q5_K_M	Tight
21.4 tok/s 13–34 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	108.8 GB	Q8_0	Tight
20.8 tok/s 12–33 · low confidence	H100 SXM5 64 GB NVIDIA	64 GB	2,020 GB/s	Mar 2023	55.9 GB	IQ4_XS	Tight
20.5 tok/s 12–33 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	108.8 GB	Q8_0	Tight
20.5 tok/s 12–33 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	108.8 GB	Q8_0	Tight
20.5 tok/s 12–33 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	85.3 GB	Q6_K	Tight
20.5 tok/s 12–33 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	85.3 GB	Q6_K	Tight
19.8 tok/s 12–32 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	61.8 GB	Q4_K_M	Tight
19.8 tok/s 12–32 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	61.8 GB	Q4_K_M	Tight
19.8 tok/s 12–32 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	61.8 GB	Q4_K_M	Tight
19.8 tok/s 12–32 · low confidence	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	61.8 GB	Q4_K_M	Tight
19.8 tok/s 12–32 · low confidence	H100 PCIe 80 GB NVIDIA	80 GB	2,040 GB/s	Oct 2022	61.8 GB	Q4_K_M	Tight
19.8 tok/s 12–32 · low confidence	H800 PCIe 80 GB NVIDIA	80 GB	2,040 GB/s	Mar 2023	61.8 GB	Q4_K_M	Tight
19.6 tok/s 12–31 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	108.8 GB	Q8_0	Comfortable
18.8 tok/s 11–30 · low confidence	A100 PCIe 80 GB NVIDIA	80 GB	1,940 GB/s	Jun 2021	61.8 GB	Q4_K_M	Tight
18.8 tok/s 11–30 · low confidence	A800 PCIe 80 GB NVIDIA	80 GB	1,940 GB/s	Nov 2022	61.8 GB	Q4_K_M	Tight
17.4 tok/s 10–28 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	108.8 GB	Q8_0	Tight
17.4 tok/s 10–28 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	108.8 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: Chinese Academy of Sciences,Harbin Institute of Technology,Nanyang Technological University,Beijing Academy of Artificial Intelligence / BAAI
Organisation type: Academia,Academia,Academia,Academia
Country: China, Singapore
Published: 7 September 2023
Authors: Xiang Li, Yiqun Yao, Xin Jiang, Xuezhi Fang, Xuying Meng, Siqi Fan, Peng Han, Jing Li, Li Du, Bowen Qin, Zheng Zhang, Aixin Sun, Yequan Wang

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, Chat

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: 101B
Training data: 311,540,000,000 tokens
Batch size: 4,310,000

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: 5.7 × 10²² 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 A800 PCIe 40 GB
Wall-clock time: 517 hours (21.5 days)

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)

How it is classified

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

Record confidence: Confident
Citations: 28

Sources

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

Reference: FLM-101B: An Open LLM and How to Train It with $100K Budget
Last updated: 25 May 2026

The extremes

The ten fastest GPUs that run FLM-101B

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 FLM-101B

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

Radeon Instinct MI200

Memory needed

55.9 GB

Fastest

33.6 tok/s

FLM-101B reaches a parameter count of 101B. That puts it above consumer hardware, into the range where a card is bought for this purpose rather than repurposed for it. The number of cards we track that can hold it: 43.

The least hardware that works is Radeon Instinct MI200, with a memory capacity of 64 GB, running it at a compression of IQ4_XS and producing around 13.2 tokens per second.

The fastest we calculate for it is B200, generating roughly 33.6 tokens per second on the strength of a memory bandwidth of 8,000 GB/s.

What this model is

FLM-101B was published by Chinese Academy of Sciences,Harbin Institute of Technology,Nanyang Technological University,Beijing Academy of Artificial Intelligence / BAAI, in the country recorded as China, during September 2023. The publishing organisation is categorised as academia,Academia,Academia,Academia.

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

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.

What decides the speed

The median result is around 18.8 tokens per second. Exceeding reading speed outright: 36 of them.

Every weight participates in every token here, so bandwidth is the whole story: the ranking below is effectively a ranking of memory throughput.

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

The training run consumed about 5.7 × 10²² FLOP, on hardware recorded as NVIDIA A800 PCIe 40 GB. That figure measures what producing the model cost, and has no bearing on how fast it answers.

The training set ran to roughly 311,540,000,000 tokens of text.

Step by step

How to choose a GPU for FLM-101B

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

Every card here has been checked against FLM-101B, needing around 55.9 GB at a compression of 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, because at long context a card that handles short questions easily can be dropped by FLM-101B.
03

Choose how far you will compress it

Each card runs the least-compressed copy it can hold, 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

Rank by throughput rather than spec sheet

Ranking by tokens per second follows memory bandwidth rather than core counts, for FLM-101B. It will not match a gaming ordering, because generation is bound by memory bandwidth. The card topping the list is B200, at 33.6 tok/s.
05

Read the fit column last

Tight means it loads and works with no room to raise the context later, in the case of FLM-101B. 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

Check the card from the other side

Each card page repeats this sweep for every model we hold, answering what else the hardware is good for beyond FLM-101B.

Answers

FLM-101B — common questions

FLM-101B— 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.

FLM-101B— how much compute was used to train it?

Training consumed around 5.7 × 10²² FLOP, on hardware recorded as NVIDIA A800 PCIe 40 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.

FLM-101B— can I run it 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 model is rarely worth using. The nearest miss we calculate falls short by 18.6 GB. Every figure here assumes the whole model is resident on the card.

FLM-101B— 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: 43. So a second card is rarely the answer here.

FLM-101B— 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: 5. Bigger cards get the more accurate version, and the quality floor above pins the comparison to one level.

FLM-101B— how accurate are these speed estimates?

Every figure is derived from memory bandwidth and model size, not benchmarked, which is why each is published as a range rather than a single number. One example: 20–54 tok/s on B200. The same model and card vary by thirty to fifty per cent depending on the inference software and its version.

FLM-101B— what GPU do I need to run it?

The smallest card in our catalogue that holds it is Radeon Instinct MI200, with a memory capacity of 64 GB. It runs the model at a compression of IQ4_XS using about 55.9 GB, and produces roughly 13.2 tokens per second. The number of cards able to run it in total: 43.

FLM-101B— how fast is it on a GPU?

It depends on the card. The quickest we calculate is B200, at about 33.6 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: 36.

FLM-101B— how much VRAM does it need?

It needs about 55.9 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.

FLM-101B— 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.

FLM-101B— how many parameters does it have?

It has a parameter count of 101B. 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.

FLM-101B— who created it?

It was published by Chinese Academy of Sciences,Harbin Institute of Technology,Nanyang Technological University,Beijing Academy of Artificial Intelligence / BAAI, based in China, an organisation categorised as academia,Academia,Academia,Academia.

FLM-101B— when was it released?

It was published in September 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.

FLM-101B— what is it used for?

It works in the domain of Language, and is recorded as handling the task of language modeling/generation, Question answering, Chat. A model can carry several of each, so these are the areas it was built for rather than a limit on what it will attempt.

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