Xinyu

Closed weights Zhejiang University (ZJU),Institute for Advanced Algorithms Research,Northeastern University (China),China Telecom,State Key Laboratory of Media Convergence Production Technology and Systems 13B parameters August 2024

No estimate

No hardware requirements for this model

The weights for this model have not been published, so it cannot be downloaded or run on your own hardware at any size. It is reachable only through its provider, and no graphics card changes that.

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
Zhejiang University (ZJU),Institute for Advanced Algorithms Research,Northeastern University (China),China Telecom,State Key Laboratory of Media Convergence Production Technology and Systems
Organisation type
Academia,Academia,Academia,Industry,Academia
Country
China
Published
23 August 2024
Authors
Yiquan Wu, Bo Tang, Chenyang Xi, Yu Yu, Pengyu Wang, Yifei Liu, Kun Kuang, Haiying Deng, Zhiyu Li, Feiyu Xiong, Jie Hu, Peng Cheng, Zhonghao Wang, Yi Wang, Yi Luo, Mingchuan Yang

What it does

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

Domain
Language
Task
Language modeling/generation
Base model
Llama 2-13B

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
13B
Training data
500,000,000 tokens

"we continued pre-training on LLaMA13B using a corpus comprising 500B tokens, which contains both English and Chinese corpus."

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.

How it was established
Hardware,Operation counting
Fine-tuning compute
2.9 × 10²² FLOP

pre-training: 312000000000000 FLOP / GPU / sec [A800 80GB reported, bf 16 assumed] * 128 GPUs * 20 days * 24 hours / day * 3600 sec / hour * 0.3 [assumed utilization] = 2.0702822e+22 FLOP SFT: 312000000000000 FLOP / GPU / sec [A800 80GB reported, bf 16 assumed] * 8 GPUs * 2 days * 24 hours / day * 3600 sec / hour * 0.3 [assumed utilization] = 1.2939264e+20 FLOP Total: 2.0832215e+22 FLOP 6 FLOP / token / parameter * 13 * 10^9 parameters * 500 * 10^9 tokens = 3.9e+22 FLOP sqrt(3.9e+22*2.083…

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 80 GB
Chips used
128
Wall-clock time
528 hours (22 days)

"The continued pre-training phase lasted 20 days on 128 Nvidia A800 80G GPUs, while the supervised fine-tuning (SFT) process took 2 days on 8 Nvidia A800 80G GPUs."

Power draw
63.1 kW

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
Closed — provider access only
Model access
Unreleased
Training code
Unreleased

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
5

Sources

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

Reference
Xinyu: An Efficient LLM-based System for Commentary Generation
Last updated
28 November 2025

What the numbers mean

What this model is

Xinyu was published by Zhejiang University (ZJU),Institute for Advanced Algorithms Research,Northeastern University (China),China Telecom,State Key Laboratory of Media Convergence Production Technology and Systems, in China, in August 2024. The organisation is categorised as academia,Academia,Academia,Industry,Academia.

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

It builds on Llama 2-13B, which is why it shares that model's general shape and size.

Because the weights are not available, none of the hardware figures elsewhere on this site apply to it.

How it was trained

It was trained on about 500,000,000 tokens of text.

Answers

Xinyu — common questions

01

Who created Xinyu?

Xinyu was published by Zhejiang University (ZJU),Institute for Advanced Algorithms Research,Northeastern University (China),China Telecom,State Key Laboratory of Media Convergence Production Technology and Systems, based in China, categorised as academia,Academia,Academia,Industry,Academia.

02

When was Xinyu released?

Xinyu was published in August 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.

03

What is Xinyu used for?

Xinyu works in Language, and is recorded as handling language modeling/generation. Models frequently carry more than one of each, and the tags describe purpose rather than capability limits.

04

What GPU do I need to run Xinyu?

None. Xinyu is a closed model — its weights were never published, so it cannot be downloaded or run on your own hardware at any price. It is reachable only through its provider.

05

Is Xinyu open source?

No. Xinyu has not had its weights published, so it exists only as a service controlled by its owner.

06

How many parameters does Xinyu have?

Xinyu has 13B 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.

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.