What are the hardware requirements for Zephyr DPO V2?

Zephyr DPO V2 requires approximately 14.4 GB of VRAM and supports a context window of 32K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed Zephyr DPO V2 and how large is it?

Zephyr DPO V2 is developed by BarraHome, a model with 7.2b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Zephyr DPO V2 perform on standard benchmarks?

Zephyr DPO V2 has the following published scores: MMLU 58.61. Compare against reference models on this page or on the LLM Explorer leaderboards.

Zephyr DPO V2 by BarraHome — VRAM 14.4GB, 32K context

Name: Zephyr DPO V2
Author: BarraHome

Zephyr DPO V2 is an open-source language model by BarraHome. Features: 7.2b LLM, VRAM: 14.4GB, Context: 32K, License: mit, Quantized, LLM Explorer Score: 0.13, Arc: 57.9, HellaSwag: 82.7, MMLU: 58.6, GSM8K: 30.3.

4bit Base model:barrahome/zephyr-dp... Base model:finetune:barrahome/... Dataset:barrahome/ultrafeedbac... Dataset:jondurbin/truthy-dpo-v... Deploy:azure En Endpoints compatible Es Mistral Model-index Quantized Region:us Safetensors Sharded Tensorflow Text-embeddings-inference Trl Unsloth

Model Card on HF 🤗: https://huggingface.co/BarraHome/zephyr-dpo-v2

Zephyr DPO V2 Parameters and Internals

Model Type

text-generation

Additional Notes

The model was developed faster using Unsloth and Huggingface's TRL library.

Supported Languages

en (fluent), es (fluent)

Training Details

Data Sources:

jondurbin/truthy-dpo-v0.1, BarraHome/ultrafeedback_binarized

Model Architecture:

Mistral

LLM Name	Zephyr DPO V2
Repository 🤗	https://huggingface.co/BarraHome/zephyr-dpo-v2
Base Model(s)	Zephyr DPO 4bit BarraHome/zephyr-dpo-4bit
Model Size	7.2b
Required VRAM	14.4 GB
Updated	2026-09-04
Maintainer	BarraHome
Model Type	mistral
Model Files	4.9 GB: 1-of-3 5.0 GB: 2-of-3 4.5 GB: 3-of-3
Supported Languages	en es
Quantization Type	4bit
Model Architecture	MistralForCausalLM
License	mit
Context Length	32768
Model Max Length	32768
Transformers Version	4.37.1
Tokenizer Class	LlamaTokenizer
Padding Token	</s>
Vocabulary Size	32000
Torch Data Type	bfloat16

Rank the Zephyr DPO V2 capabilities

Have you tried this model? Rate its performance — this feedback helps the ML community find the right model for their needs.

Instruction Following and Task Automation
Factuality and Completeness of Knowledge
Censorship and Alignment
Data Analysis and Insight Generation
Text Generation
Text Summarization and Feature Extraction
Code Generation
Multi-Language Support and Translation

Best Alternatives to Zephyr DPO V2

Best Alternatives	Context / RAM	Downloads
...Sft Bnb 4bit DPO Mtbr 180steps	32K / 14.4 GB	5
...Sft Bnb 4bit DPO Mtbc 213steps	32K / 14.4 GB	5
...Sft Bnb 4bit DPO Mtbo 180steps	32K / 14.4 GB	6
Mixtral V0.3 Full 16bit V2	32K / 14.5 GB	6
...88 07.02 RP DPO Merged 16bit 3	32K / 14.4 GB	25
Rezephyr DPO	32K / 14.4 GB	20
Autogen Mistral 16bit Merged	32K / 14.5 GB	143
Tools Mistral 16bit Merged New	32K / 14.5 GB	5
AI Tutor	32K / 14.5 GB	6
...Web AI HumanAI 012 INSTRUCT XA	512K / 14.4 GB	9

Note: green Score (e.g. "73.2") means that the model is better than BarraHome/zephyr-dpo-v2.