Telechat2-115B TPS calculator

Open weights China Telecom 115B parameters September 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

43 of 818 cards that can run it

Smallest card that fits

Radeon Instinct MI200

64 GB · Q3_K_M · 12.7 tok/s

Fastest card

B200

29.5 tok/s · 180 GB

Which GPUs can run Telechat2-115B?

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
29.5 tok/s 18–47 · low confidence	B200 NVIDIA	180 GB	8,000 GB/s	Jan 2024	123.8 GB	Q8_0	Comfortable
29.5 tok/s 18–47 · low confidence	B300 NVIDIA	288 GB	8,000 GB/s	Sep 2025	123.8 GB	Q8_0	Comfortable
28.6 tok/s 17–46 · low confidence	H100 SXM5 80 GB NVIDIA	80 GB	3,360 GB/s	Oct 2022	70.3 GB	Q4_K_M	Tight
28.6 tok/s 17–46 · low confidence	H800 SXM5 NVIDIA	80 GB	3,360 GB/s	Mar 2023	70.3 GB	Q4_K_M	Tight
27.3 tok/s 16–44 · low confidence	Radeon Instinct MI300 AMD	128 GB	6,550 GB/s	Jan 2023	97.0 GB	Q6_K	Tight
25.9 tok/s 16–41 · low confidence	H100 NVL 94 GB NVIDIA	94 GB	3,940 GB/s	Mar 2023	83.7 GB	Q5_K_M	Tight
23.5 tok/s 14–38 · low confidence	Radeon Instinct MI350X AMD	288 GB	8,190 GB/s	Jan 2025	123.8 GB	Q8_0	Comfortable
23.5 tok/s 14–38 · low confidence	Radeon Instinct MI355X AMD	288 GB	8,190 GB/s	Jan 2025	123.8 GB	Q8_0	Comfortable
22.2 tok/s 13–36 · low confidence	Radeon Instinct MI300A AMD	128 GB	5,325 GB/s	Dec 2023	97.0 GB	Q6_K	Tight
22.1 tok/s 13–35 · low confidence	H100 PCIe 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	83.7 GB	Q5_K_M	Tight
22.1 tok/s 13–35 · low confidence	H100 SXM5 94 GB NVIDIA	94 GB	3,360 GB/s	Mar 2023	83.7 GB	Q5_K_M	Tight
22.1 tok/s 13–35 · low confidence	H100 SXM5 96 GB NVIDIA	96 GB	3,360 GB/s	Mar 2023	83.7 GB	Q5_K_M	Tight
20.1 tok/s 12–32 · low confidence	H100 SXM5 64 GB NVIDIA	64 GB	2,020 GB/s	Mar 2023	56.9 GB	Q3_K_M	Tight
18.0 tok/s 11–29 · low confidence	H200 NVL NVIDIA	141 GB	4,890 GB/s	Nov 2024	123.8 GB	Q8_0	Tight
18.0 tok/s 11–29 · low confidence	H200 SXM 141 GB NVIDIA	141 GB	4,890 GB/s	Nov 2024	123.8 GB	Q8_0	Tight
17.3 tok/s 10–28 · low confidence	A100 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Nov 2020	70.3 GB	Q4_K_M	Tight
17.3 tok/s 10–28 · low confidence	A100X NVIDIA	80 GB	2,040 GB/s	Jun 2021	70.3 GB	Q4_K_M	Tight
17.3 tok/s 10–28 · low confidence	A800 SXM4 80 GB NVIDIA	80 GB	2,040 GB/s	Aug 2022	70.3 GB	Q4_K_M	Tight
17.3 tok/s 10–28 · low confidence	H100 CNX NVIDIA	80 GB	2,040 GB/s	Mar 2023	70.3 GB	Q4_K_M	Tight
17.3 tok/s 10–28 · low confidence	H100 PCIe 80 GB NVIDIA	80 GB	2,040 GB/s	Oct 2022	70.3 GB	Q4_K_M	Tight
17.3 tok/s 10–28 · low confidence	H800 PCIe 80 GB NVIDIA	80 GB	2,040 GB/s	Mar 2023	70.3 GB	Q4_K_M	Tight
17.2 tok/s 10–28 · low confidence	Radeon Instinct MI325X AMD	256 GB	6,000 GB/s	Oct 2024	123.8 GB	Q8_0	Comfortable
16.5 tok/s 10–26 · low confidence	A100 PCIe 80 GB NVIDIA	80 GB	1,940 GB/s	Jun 2021	70.3 GB	Q4_K_M	Tight
16.5 tok/s 10–26 · low confidence	A800 PCIe 80 GB NVIDIA	80 GB	1,940 GB/s	Nov 2022	70.3 GB	Q4_K_M	Tight
15.3 tok/s 9–24 · low confidence	Radeon Instinct MI300X AMD	192 GB	5,325 GB/s	Dec 2023	123.8 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: China Telecom
Organisation type: Industry
Country: China
Published: 20 September 2024
Authors: Zihan Wang and Xinzhang Liu and Shixuan Liu and Yitong Yao and Yuyao Huang and Zhongjiang He and Xuelong Li and Yongxiang Li and Zhonghao Che and Zhaoxi Zhang and Yan Wang and Xin Wang and Luwen Pu and Huihan Xu and Ruiyu Fang and Yu Zhao and Jie Zhang and Xiaomeng Huang and Zhilong Lu and Jiaxin Peng and Wenjun Zheng and Shiquan Wang and Bingkai Yang and Xuewei he and Zhuoru Jiang and Qiyi Xie an…

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, Chat, Text summarization, Code generation
Approach: Supervised
Numerical format: FP16

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: 115B
Training data: 10,000,000,000,000 tokens
Epochs: 1
Batch size: 6,000,000

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: 6.9 × 10²⁴ 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.

Data centre: There is no paper to reference, no information about hardware used for training found in media.

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: Open (restricted use)
Hugging Face: Tele-AI

How it is classified

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

Likely above 10²³ FLOP: Yes
Why it is tracked: Training cost
Record confidence: Confident

Sources

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

Reference: TeleChat Technical Report
Last updated: 3 January 2026

The extremes

The ten fastest GPUs for Telechat2-115B

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 Telechat2-115B

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 it takes to run this model

Minimum card

Radeon Instinct MI200

Memory needed

56.9 GB

Fastest

29.5 tok/s

Telechat2-115B sits at 115B parameters, which puts it above consumer hardware and into the range where a card is bought for this purpose rather than repurposed for it. 43 of the cards we track can hold it.

The smallest card that holds it is the Radeon Instinct MI200 with 64 GB, running it at Q3_K_M and producing around 12.7 tokens per second.

At the other end, a B200 generates roughly 29.5 tokens per second on it, on the strength of 8,000 GB/s of memory bandwidth.

What this model is

Telechat2-115B was published by China Telecom, in China, in September 2024. It comes out of industry.

It works in Language, and is recorded as doing language modeling/generation, Question answering, Quantitative reasoning, Chat, Text summarization, Code generation.

Because its weights were released, nothing about running it depends on a provider staying available — it is yours once downloaded. It is published under the Tele-AI organisation on Hugging Face.

What decides the speed

Across every card that can run it, the middle of the range is about 17.2 tokens per second, and 39 of them clear the ten tokens per second that roughly matches reading speed.

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.

Its internal architecture is not on file, so memory is approximated from the parameter count and marked accordingly. Expect the real figure to differ, more so at long context.

What went into building it

Training it took roughly 6.9 × 10²⁴ FLOP of computation — a measure of what producing the model cost, not of how fast it answers.

Around 10,000,000,000,000 tokens went into training it.

The reason it appears in this catalogue at all is training cost.

Step by step

How to choose a GPU for Telechat2-115B

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 Telechat2-115B — around 56.9 GB at Q3_K_M. Capacity is the gate — a card either holds it or it does not.
02

Set the context length you will work at

Set the context to what you will actually use. The cache grows with the conversation, and it is the usual reason Telechat2-115B stops fitting a card that seemed fine.
03

Decide how much compression you will accept

The quantisation column varies by card, because a bigger card holds a more accurate copy of Telechat2-115B — Q3_K_M on the smallest card that fits. Set a floor to hold the comparison at one level.
04

Sort by speed

Sort by speed to see how cards rank for Telechat2-115B. It will not match a gaming ordering — generation is bound by memory bandwidth, which is why the B200 tops it at 29.5 tok/s.
05

Look at the headroom, not just the fit

A tight fit runs Telechat2-115B but leaves nothing spare for a longer conversation; comfortable has headroom. If you expect to grow the context, buy for comfortable.
06

See what else that card runs

Following a card through to its own page shows every other model it can hold, which is the question that follows once Telechat2-115B is settled.

Answers

Telechat2-115B — common questions

Where can I download Telechat2-115B?

Its weights are published under the Tele-AI 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 Telechat2-115B?

Around 6.9 × 10²⁴ 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 Telechat2-115B 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 27.1 GB. Our figures for Telechat2-115B assume it is fully resident.

Would two GPUs run Telechat2-115B faster?

A second card roughly doubles the memory available but not the generation rate. With 43 cards already able to run Telechat2-115B alone, the case for pairing is weak.

Why does the quantisation differ between cards for Telechat2-115B?

A larger card holds a more accurate copy. Across the cards that run Telechat2-115B, 6 compression levels are used; the floor control above pins it to one.

How accurate are these Telechat2-115B speed estimates?

They are calculated from specifications rather than measured, and each carries a range — 18–47 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 Telechat2-115B?

The smallest card in our catalogue that holds Telechat2-115B is the Radeon Instinct MI200, with 64 GB of memory. It runs the model at Q3_K_M using about 56.9 GB, and produces roughly 12.7 tokens per second. 43 cards in total can run it.

How fast is Telechat2-115B on a GPU?

It depends on the card. The quickest we calculate is a B200 at about 29.5 tokens per second; the slowest that still runs it manages considerably less. Reading speed is around ten tokens per second, and 39 of the cards that can run Telechat2-115B clear that.

How much VRAM does Telechat2-115B need?

About 56.9 GB at Q3_K_M 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.

Is Telechat2-115B open source?

Its weights are published, so Telechat2-115B 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 Telechat2-115B have?

Telechat2-115B has 115B parameters. 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 Telechat2-115B?

Telechat2-115B was published by China Telecom, based in China, categorised as industry.

When was Telechat2-115B released?

Telechat2-115B was published in September 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.

What is Telechat2-115B used for?

Telechat2-115B works in Language, and is recorded as handling language modeling/generation, Question answering, Quantitative reasoning, Chat, Text summarization, Code generation. 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 3 January 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