What are the hardware requirements for SUS Chat 34B?

SUS Chat 34B requires approximately 68.9 GB of VRAM and supports a context window of 8K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed SUS Chat 34B and how large is it?

SUS Chat 34B is developed by sustech, a model with 34b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does SUS Chat 34B perform on standard benchmarks?

SUS Chat 34B has the following published scores: MMLU 75.9. Compare against reference models on this page or on the LLM Explorer leaderboards.

SUS Chat 34B by sustech — VRAM 68.9GB, 8K context

Name: SUS Chat 34B
Author: sustech

SUS Chat 34B is an open-source language model by sustech. Features: 34b LLM, VRAM: 68.9GB, Context: 8K, License: apache-2.0, HF Score: 73.2, LLM Explorer Score: 0.13, Arc: 97.7, HellaSwag: 94.2, MMLU: 75.9, TruthfulQA: 57, WinoGrande: 83.5, GSM8K: 72.3.

Doi:10.57967/hf/1718 Endpoints compatible Llama Pytorch Region:us Sharded

Model Card on HF 🤗: https://huggingface.co/SUSTech/SUS-Chat-34B

SUS Chat 34B Benchmarks

ARC: 97.7 vs 96.7 (so35)^1%

HellaSwag: 94.2 vs 95.3 (gpt4)^-1.2%

MMLU: 75.9 vs 88.3 (so35)^-14%

TruthfulQA: 57.04 vs 59 (gpt4)^-3.3%

WinoGrande: 83.5 vs 87.5 (gpt4)^-4.6%

GSM8K: 72.3 vs 96.4 (so35)^-25%

OpenCompas: 64.3 vs 73.3 (gpt4)^-12.3%

Exam: 81.5 vs 77.2 (gpt4)^5.6%

Language: 38.2 vs 62 (gpt4)^-38.4%

Knowledge: 62.7 vs 73.5 (gpt4)^-14.7%

Understand: 74.1 vs 70 (gpt4)^5.9%

Reasoning: 63.9 vs 74.4 (gpt4)^-14.1%

LLME Score: 0.12933

^nn.n% — How the model compares to the reference models: Anthropic Sonnet 3.5 ("so35"), GPT-4o ("gpt4o") or GPT-4 ("gpt4").

What is the LLM Explorer Rank (Score)

🌟 Advertise your project 🚀

SUS Chat 34B Parameters and Internals

Model Type

bilingual Chinese-English, dialogue model, instruction-tuned

Use Cases

Areas:

research, commercial applications

Applications:

multilingual tasks, long-text dialogues, complex instructions

Primary Use Cases:

multi-turn dialogues, instruction following

Supported Languages

Chinese (high proficiency), English (high proficiency)

Training Details

Data Volume:

1.4 billion tokens

Methodology:

fine-tuning

Context Length:

8000

Model Architecture:

based on LLaMA architecture

Release Notes

Version:

2023-12-01

Date:

2023-12-01

Notes:

Release of SUS-Chat-34B model, a 34B bilingual Chinese-English dialogue model.

LLM Name	SUS Chat 34B
Repository 🤗	https://huggingface.co/SUSTech/SUS-Chat-34B
Model Size	34b
Required VRAM	68.9 GB
Updated	2026-05-14
Maintainer	sustech
Model Type	llama
Model Files	10.0 GB: 1-of-7 9.9 GB: 2-of-7 9.8 GB: 3-of-7 9.8 GB: 4-of-7 9.8 GB: 5-of-7 9.9 GB: 6-of-7 9.7 GB: 7-of-7
Model Architecture	LlamaForCausalLM
License	apache-2.0
Context Length	8192
Model Max Length	8192
Transformers Version	4.35.0
Tokenizer Class	LlamaTokenizer
Padding Token	<unk>
Vocabulary Size	64000
Torch Data Type	bfloat16

Quantized Models of the SUS Chat 34B

Model	Likes	Downloads	VRAM
SUS Chat 34B GGUF	15	387	14 GB
SUS Chat 34B AWQ	9	12	19 GB
SUS Chat 34B GPTQ	3	4	18 GB

Best Alternatives to SUS Chat 34B

Best Alternatives	Context / RAM	Downloads	Likes
Bagel Hermes 34B Slerp	195K / 68.9 GB	8371	1
34B Beta	195K / 69.2 GB	8242	66
Smaug 34B V0.1	195K / 69.2 GB	8419	64
Yi 34B 200K	195K / 68.9 GB	10338	320
Casual Magnum 34B	195K / 68.8 GB	9	1
Bagel 34B V0.2	195K / 68.7 GB	401	41
Yi 34B 200K AEZAKMI V2	195K / 69.2 GB	537	12
Smaug 34B V0.1 ExPO	195K / 69.2 GB	7738	0
Bagel DPO 34B V0.5	195K / 68.7 GB	8205	17
Faro Yi 34B	195K / 69.2 GB	8432	6

Note: green Score (e.g. "73.2") means that the model is better than SUSTech/SUS-Chat-34B.

Rank the SUS Chat 34B 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? 53764 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

SUS Chat 34B by sustech