What are the hardware requirements for 34B Beta 5.0bpw H6 EXL2?

34B Beta 5.0bpw H6 EXL2 requires approximately 22.3 GB of VRAM and supports a context window of 195K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed 34B Beta 5.0bpw H6 EXL2 and how large is it?

34B Beta 5.0bpw H6 EXL2 is developed by LoneStriker, a model with 34b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

Where can I download or evaluate 34B Beta 5.0bpw H6 EXL2?

34B Beta 5.0bpw H6 EXL2 is hosted on Hugging Face and linked from this page. LLM Explorer also lists quantized variants and similar alternatives if available.

34B Beta 5.0bpw H6 EXL2 by LoneStriker — VRAM 22.3GB, 195K context

Name: 34B Beta 5.0bpw H6 EXL2
Author: LoneStriker

34B Beta 5.0bpw H6 EXL2 is an open-source language model by LoneStriker. Features: 34b LLM, VRAM: 22.3GB, Context: 195K, License: gpl-3.0, Quantized, LLM Explorer Score: 0.11.

Conversational Endpoints compatible Exl2 Llama Quantized Region:us Safetensors Sharded Tensorflow

Model Card on HF 🤗: https://huggingface.co/LoneStriker/34b-beta-5.0bpw-h6-exl2

34B Beta 5.0bpw H6 EXL2 Parameters and Internals

Model Type

Causal Language Model

Use Cases

Considerations:

Avoid using "accelerated inference frameworks" like VLLM due to precision issues leading to significant output quality degradation. Use Transformers for inference, or consider using q8_0 quantization with llama.cpp temporarily for faster inference.

Additional Notes

Some issues with model weights in terms of precision, to be fixed in the next version update. Repetition penalty should not be used. Re-alignment on synthetic datasets has changed distribution significantly from original datasets.

LLM Name	34B Beta 5.0bpw H6 EXL2
Repository 🤗	https://huggingface.co/LoneStriker/34b-beta-5.0bpw-h6-exl2
Model Size	34b
Required VRAM	22.3 GB
Updated	2026-08-07
Maintainer	LoneStriker
Model Type	llama
Model Files	8.6 GB: 1-of-3 8.6 GB: 2-of-3 5.1 GB: 3-of-3
Quantization Type	exl2
Model Architecture	LlamaForCausalLM
License	gpl-3.0
Context Length	200000
Model Max Length	200000
Transformers Version	4.37.2
Tokenizer Class	LlamaTokenizer
Padding Token	<unk>
Vocabulary Size	64000
Torch Data Type	bfloat16

Rank the 34B Beta 5.0bpw H6 EXL2 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 34B Beta 5.0bpw H6 EXL2

Best Alternatives	Context / RAM	Downloads	Likes
...34B 200K Aezakmi Raw 1902 EXL2	195K / 20.7 GB	3	1
Yi 34B 200K MAR2024 EXL2 4bpw	195K / 18 GB	4	1
...B 200K RPMerge 4.65bpw H6 EXL2	195K / 10.5 GB	8	2
...mi Raw Toxic 2702 4.65bpw EXL2	195K / 20.8 GB	7	2
Yi 34B 200K RPMerge	195K / 68.9 GB	72	60
Opus V1 34B 5.0bpw H6 EXL2	195K / 22.3 GB	5	2
Opus V1 34B 8.0bpw H8 EXL2	195K / 34.8 GB	7	1
Opus V1 34B 4.0bpw H6 EXL2	195K / 18 GB	5	1
34B Beta 4.0bpw H6 EXL2	195K / 18.1 GB	2	1
Smaug 34B V0.1 2.65bpw H6 EXL2	195K / 12.4 GB	4	1

Note: green Score (e.g. "73.2") means that the model is better than LoneStriker/34b-beta-5.0bpw-h6-exl2.