Text

Zephyr

Hugging Face educational chat models based on Mistral, Gemma and Mixtral with an open fine-tuning recipe. Zephyr 7B Beta showed a small model can be trained to large-model level without human labeling.

The last open version came out in Apr 2024. The family has not been updated for a long time: the model still works, but do not expect fixes or new sizes.

Developer
Hugging Face (H4), USA
First release
Oct 2023
Latest release
Apr 2024
Sizes
7B – 141B-A35B
License
Commercial use allowedMIT (Zephyr 7B), Apache 2.0 (Zephyr 141B); the Gemma-based version: Gemma Terms
Russian
Not supported
Ready-made builds
GGUF, GPTQ
Running
Available in OllamaNeeds a GPU
Industries
Customer support, Science and research
Deploy on your server Hugging Face GitHub

What it does

  • Lightweight chat assistant
  • Reference and starting point for your own fine-tuning
  • Drafts of texts and replies

Where it is used

IT and R&DEducationSmall business

Hardware requirements

LaptopLaptop or regular PC, up to 8 GB of VRAM — smaller versions
fits
1 GPUOne GPU with 16–80 GB — mid-size versions
fits
ClusterServer with several GPUs — flagship versions
fits

Versions

  1. Zephyr ORPO 141B-A35B (Mixtral)
  2. Zephyr 7B Gemma
  3. Zephyr 7B Beta
  4. Zephyr 7B Alpha

How to run it

With Ollama — the fastest wayInstall Ollama on your computer or server and find the model in its library by name. Good enough to try the model on your own tasks in one evening.Find it in the Ollama library
On your own serverWeights are downloaded from Hugging Face and served with vLLM (load and API) or llama.cpp (modest hardware). The model then runs in a closed network with no per-request fees.Hugging Face
How much hardware you needCalculate the VRAM for the model size, context length and number of concurrent requests.Open the hardware calculator

I can set this up end to end: pick the model size, deploy it on your server and connect it to your systems. Quantization compresses a model so it takes less video memory and runs on more modest hardware. Answers change slightly, so quality is checked on your own examples.

Frequently asked questions

Can Zephyr be used in a commercial project?

Yes. License: MIT (Zephyr 7B), Apache 2.0 (Zephyr 141B); the Gemma-based version: Gemma Terms. It allows commercial use, but it is still worth having a lawyer review the license before launch.

What hardware does Zephyr need?

At minimum: Laptop or regular PC, up to 8 GB of VRAM — smaller versions. Without a GPU the model is not practical. You can calculate the exact VRAM for your model size and context in the hardware calculator.

Does Zephyr support Russian?

No. The model card lists its languages and Russian is not among them.

Where can I download Zephyr and what does it cost?

The Zephyr weights are open and free to download. You only pay for the hardware it runs on and for the setup. Source links are at the bottom of this page.

How I deploy it for clients

  1. SelectionI pick the model size for your task and hardware and test it on your examples.
  2. DeploymentI deploy it on your server or in a closed network and provide an API.
  3. Fine-tuningI fine-tune it on your data (LoRA) or connect a knowledge base — whichever is cheaper for the task.
  4. IntegrationI connect it to your CRM, ERP, bot, website or team chat and set up monitoring.

Similar models

TextOpenChatOpenChat (Tsinghua University) · ChinaCommercial use with conditions

Fine-tunes of Mistral 7B and Llama 3 8B using the C-RLFT method that caught up with ChatGPT-3.5 in 2023–2024 at just 7–8B. A lightweight general-purpose assistant for a modest server.

DetailsTextMistralMistral AI · FranceCommercial use with conditions

European models focused on speed. Mixtral was one of the first open mixture-of-experts models; there are versions for images (Pixtral, Medium 3.5), Lean proofs and moderation (Shieldstral).

DetailsTextTuluAi2 · USACommercial use with conditions

Ai2 fine-tunes of Llama with a fully open recipe: data, code and all intermediate stages. Tulu 3 405B is one of the largest openly fine-tuned models; OLMo chat versions use the same recipe.

Details

Source: huggingface.co/HuggingFaceH4/zephyr-orpo-141b-A35b-v0.1. Data checked against the model card on 22 Sep 2026. Have a lawyer review the license before commercial launch.