LLM EXPLORER
 LLM Explorer: A Curated Large Language Model Directory and Analytics
Dark Theme
  Agents   Benchmarks   LLM  List    LLM Hosting    LLM Leaderboards    Blog    Newsfeed    Advertise

Beyonder 4x7B V3 by mlabonne

 »  All LLMs  »  mlabonne  »  Beyonder 4x7B V3   URL Share it on

  7856   

Beyonder 4x7B V3 is an open-source language model by mlabonne. Features: 24.2b LLM, VRAM: 48.3GB, Context: 32K, License: cc-by-nc-4.0, MoE, HF Score: 75.7, LLM Explorer Score: 0.22, Arc: 71.7, HellaSwag: 88.9, MMLU: 64.9, TruthfulQA: 74.4, WinoGrande: 83.5, GSM8K: 70.6.

Base model:beowolx/codeninja-1... Base model:merge:beowolx/coden... Base model:merge:mlabonne/alph... Base model:merge:mlabonne/neur... Base model:merge:sanjiwatsuki/... Base model:mlabonne/alphamonar... Base model:mlabonne/neuraldare... Base model:sanjiwatsuki/kunoic...   Conversational   Deploy:azure   Endpoints compatible   Frankenmoe   Lazymergekit   Merge   Mergekit   Mixtral   Moe   Region:us   Safetensors   Sharded   Tensorflow
Model Card on HF 🤗: https://huggingface.co/mlabonne/Beyonder-4x7B-v3 

Beyonder 4x7B V3 Benchmarks

MMLU Pro: 16.81
GPQA: 4.7
MUSR: 8.93
BBH: 24.56
IFEval: 56.08 vs 88 (so35)-36.3%
ARC: 71.67 vs 96.7 (so35)-25.9%
HellaSwag: 88.86 vs 95.3 (gpt4)-6.8%
MMLU: 64.87 vs 88.3 (so35)-26.5%
TruthfulQA: 74.44 vs 59 (gpt4)26.2%
WinoGrande: 83.5 vs 87.5 (gpt4)-4.6%
GSM8K: 70.58 vs 96.4 (so35)-26.8%
MATH Lvl 5: 5.36
LLME Score: 0.22165
nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").
Beyonder 4x7B V3 (mlabonne/Beyonder-4x7B-v3)
🌟 Advertise your project 🚀

Beyonder 4x7B V3 Parameters and Internals

Model Type 
Mixture of Experts (MoE), text-generation
Use Cases 
Applications:
chat, programming, storytelling, mathematics
Additional Notes 
The model leverages four experts, each specializing in specific task types for well-rounded capabilities.
Input Output 
Input Format:
Mistral Instruct chat template
Accepted Modalities:
text
Performance Tips:
Use Mistral Instruct chat template for optimal results, tweak inference parameters if using SillyTavern.
LLM NameBeyonder 4x7B V3
Repository 🤗https://huggingface.co/mlabonne/Beyonder-4x7B-v3 
Base Model(s)  AlphaMonarch 7B   beowolx/CodeNinja-1.0-OpenChat-7B   Kunoichi DPO V2 7B   NeuralDaredevil 7B   mlabonne/AlphaMonarch-7B   beowolx/CodeNinja-1.0-OpenChat-7B   SanjiWatsuki/Kunoichi-DPO-v2-7B   mlabonne/NeuralDaredevil-7B
Model Size24.2b
Required VRAM48.3 GB
Updated2026-03-31
Maintainermlabonne
Model Typemixtral
Model Files  9.9 GB: 1-of-5   9.9 GB: 2-of-5   10.0 GB: 3-of-5   10.0 GB: 4-of-5   8.5 GB: 5-of-5
Model ArchitectureMixtralForCausalLM
Licensecc-by-nc-4.0
Context Length32768
Model Max Length32768
Transformers Version4.39.0
Tokenizer ClassLlamaTokenizer
Padding Token<s>
Vocabulary Size32000
Torch Data Typefloat16

Best Alternatives to Beyonder 4x7B V3

Best Alternatives
Context / RAM
Downloads
Likes
Dzakwan MoE 4x7b Beta32K / 48.4 GB90720
Calme 4x7B MoE V0.232K / 48.3 GB80192
Calme 4x7B MoE V0.132K / 48.3 GB79052
Mera Mix 4x7B32K / 48.3 GB815719
MixtureofMerges MoE 4x7b V532K / 48.3 GB78111
MixtureofMerges MoE 4x7b V432K / 48.3 GB78894
CognitiveFusion2 4x7B BF1632K / 48.3 GB80763
Proto Athena 4x7B32K / 48.4 GB50
Proto Athena V0.2 4x7B32K / 48.4 GB80
NeuralStar FusionWriter 4x7b32K / 48.3 GB249
Note: green Score (e.g. "73.2") means that the model is better than mlabonne/Beyonder-4x7B-v3.

Rank the Beyonder 4x7B V3 Capabilities

🆘 Have you tried this model? Rate its performance. This feedback would greatly assist ML community in identifying the most suitable model for their needs. Your contribution really does make a difference! 🌟

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  

What open-source LLMs or SLMs are you in search of? 53254 in total.

Email us: info@extractum.io. Our Privacy Policy  | Terms and Conditions  | Suggest an improvement.
Our Social Media →  
Original data from HuggingFace, OpenCompass and various public git repos.
Check out Ag3ntum — our secure, self-hosted AI agent for server management.
Release v20260328a
  Support LLM Explorer