Step-Omni
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
- StepFun
- Organisation type
- Industry
- Country
- China
- Published
- 18 February 2025
- Authors
- Ailin Huang, Boyong Wu, Bruce Wang, Chao Yan, Chen Hu, Chengli Feng, Fei Tian, Feiyu Shen, Jingbei Li, Mingrui Chen, Peng Liu, Ruihang Miao, Wang You, Xi Chen, Xuerui Yang, Yechang Huang, Yuxiang Zhang, Zheng Gong, Zixin Zhang, Hongyu Zhou, Jianjian Sun, Brian Li, Chengting Feng, Changyi Wan, Hanpeng Hu, Jianchang Wu, Jiangjie Zhen, Ranchen Ming, Song Yuan, Xuelin Zhang, Yu Zhou, Bingxin Li, Buyun…
What it does
The problem areas the model was built for. A model can carry several of each.
- Domain
- Speech, Language, Multimodal, Vision
- Task
- Speech synthesis, Speech recognition (ASR), Speech-to-text, Text-to-speech (TTS), Audio question answering, Audio generation, Image captioning, Visual question answering, Speech-to-speech
- Base model
- Step-1
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
- 130B
- Training data
- 2,468,000,000,000 tokens
130B
1.1 trillion tokens of audio continuation data (approximately 7,300,000 hours) + 113 billion tokens of TTS (about 700,000 hours) + 105 billion tokens of ASR data (around 650,000 hours) + 350 billion tokens of audio-text alternating data (approximately 2,000,000 hours) + 800 billion tokens of tokens of image-text paired/alternating data = 2468 billion tokens
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
- 2.5 × 10²⁴ FLOP
- How it was established
- Operation counting
base LLM 6.24e+23 FLOP + 6 FLOP / parameter / token * 130*10^9 parameters * 2468000000000 tokens [see training dataset size notes] = 6.24e+23 FLOP + 1.92504e+24 FLOP = 2.54904e+24 FLOP
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 H800 SXM5
- Hardware utilisation
- MFU 35.0%
"We train Step-Omni on thousands of H800 GPUs with 35% Model Flops Utilization (MFU)."
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
Sources
Where this record came from and when it was last checked.
- Reference
- Step-Audio: Unified Understanding and Generation in Intelligent Speech Interaction
- Last updated
- 28 November 2025
What the numbers mean
Background
Step-Omni was published by StepFun, in China, in February 2025. It comes out of industry.
It works in Speech, Language, Multimodal, Vision, and is recorded as doing speech synthesis, Speech recognition (ASR), Speech-to-text, Text-to-speech (TTS), Audio question answering, Audio generation, Image captioning, Visual question answering, Speech-to-speech.
Its starting point was Step-1 — most models at this scale are adapted from an existing base rather than built from nothing.
Because the weights are not available, none of the hardware figures elsewhere on this site apply to it.
What went into building it
The training run consumed about 2.5 × 10²⁴ FLOP, on NVIDIA H800 SXM5. That figure describes the cost of creating it and has no bearing on how quickly it generates text.
The training set ran to roughly 2,468,000,000,000 tokens.
Answers
Step-Omni — common questions
What is Step-Omni used for?
Step-Omni works in Speech, Language, Multimodal, Vision, and is recorded as handling speech synthesis, Speech recognition (ASR), Speech-to-text, Text-to-speech (TTS), Audio question answering, Audio generation, Image captioning, Visual question answering, Speech-to-speech. These are the areas it was designed around; they describe intent rather than a hard boundary.
How much compute was used to train Step-Omni?
Around 2.5 × 10²⁴ FLOP, on NVIDIA H800 SXM5. 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.
What GPU do I need to run Step-Omni?
None. Step-Omni 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.
Is Step-Omni open source?
No. Step-Omni has not had its weights published, so it exists only as a service controlled by its owner.
How many parameters does Step-Omni have?
Step-Omni has 130B parameters. 130B. 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 Step-Omni?
Step-Omni was published by StepFun, based in China, categorised as industry.
When was Step-Omni released?
Step-Omni was published in February 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.