DreamLLM

Closed weights Xi’an Jiaotong University,Megvii Inc,Tsinghua University,Huazhong University of Science and Technology 7B parameters September 2023

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
Xi’an Jiaotong University,Megvii Inc,Tsinghua University,Huazhong University of Science and Technology
Organisation type
Academia,Industry,Academia,Academia
Country
China
Published
20 September 2023
Authors
Runpei Dong, Chunrui Han, Yuang Peng, Zekun Qi, Zheng Ge, Jinrong Yang, Liang Zhao, Jianjian Sun, Hongyu Zhou, Haoran Wei, Xiangwen Kong, Xiangyu Zhang, Kaisheng Ma, Li Yi

What it does

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

Domain
Multimodal, Language, Vision, Image generation
Task
Language modeling/generation, Vision-language generation, Image generation

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
7B

7B

Training data
70,369,280,000 tokens

from Table 13: 30M pairs image and text, 4M interleaved image-text documents, and 120k instruction examples

Epochs
1

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
7.5 × 10²⁰ FLOP

Table 13: 128xA800 GPU for (6+10+1.5) hours flops = (128) * ( 312 * 10**12) * (17.5 * 3600) * (0.3) = 7.5e20 (num gpu) * (peak flops) * (time in seconds) * (assumed utilization rate)

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
Chips used
128
Chip-hours
2,240
Wall-clock time
18 hours

from Table 11: (6+10+1.5) hours

Power draw
63.6 kW

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
318

Sources

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

Reference
DreamLLM: Synergistic Multimodal Comprehension and Creation
Last updated
25 May 2026

What the numbers mean

About this model

DreamLLM was published by Xi’an Jiaotong University,Megvii Inc,Tsinghua University,Huazhong University of Science and Technology, in China, in September 2023. It comes out of academia,Industry,Academia,Academia.

It works in Multimodal, Language, Vision, Image generation, and is recorded as doing language modeling/generation, Vision-language generation, Image generation.

This is a closed model: the trained values stayed with whoever produced them, and there is no local version to run.

What went into building it

Producing it required around 7.5 × 10²⁰ FLOP of arithmetic, on NVIDIA A800 PCIe 40 GB, which is a statement about the training budget rather than about inference.

Around 70,369,280,000 tokens went into training it.

Answers

DreamLLM — common questions

01

Is DreamLLM open source?

The licensing for DreamLLM was never recorded in our source data. We treat unstated licensing as closed, because an unrecorded licence is not one to rely on.

02

How many parameters does DreamLLM have?

DreamLLM has 7B parameters. 7B. 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.

03

Who created DreamLLM?

DreamLLM was published by Xi’an Jiaotong University,Megvii Inc,Tsinghua University,Huazhong University of Science and Technology, based in China, categorised as academia,Industry,Academia,Academia.

04

When was DreamLLM released?

DreamLLM 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.

05

What is DreamLLM used for?

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

06

How much compute was used to train DreamLLM?

Around 7.5 × 10²⁰ FLOP, on 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.

07

What GPU do I need to run DreamLLM?

None. DreamLLM 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.

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.