What are the hardware requirements for OPT 30B Erebus?

OPT 30B Erebus requires approximately 36 GB of VRAM and supports a context window of 2K tokens. Quantized variants may run on less VRAM; see the Quantized Models section on this page.

Who developed OPT 30B Erebus and how large is it?

OPT 30B Erebus is developed by KoboldAI, a model with 30b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does OPT 30B Erebus perform on standard benchmarks?

OPT 30B Erebus has the following published scores: MMLU 24.8. Compare against reference models on this page or on the LLM Explorer leaderboards.

OPT 30B Erebus by KoboldAI — VRAM 36GB, 2K context

Name: OPT 30B Erebus
Author: KoboldAI

OPT 30B Erebus is an open-source language model by KoboldAI. Features: 30b LLM, VRAM: 36GB, Context: 2K, License: other, LLM Explorer Score: 0.09, Arc: 36.7, HellaSwag: 65.6, MMLU: 24.8, GSM8K: 0.2.

Arxiv:2205.01068 En Opt Pytorch Region:us Sharded

Model Card on HF 🤗: https://huggingface.co/KoboldAI/OPT-30B-Erebus

OPT 30B Erebus Parameters and Internals

Model Type

text generation

Additional Notes

Warning: THIS model has a very strong NSFW bias.

Training Details

Data Sources:

Literotica (everything with 4.5/5 or higher), Sexstories (everything with 90 or higher), Dataset-G (private dataset of X-rated stories), Doc's Lab (all stories), Pike Dataset (novels with 'adult' rating), SoFurry (collection of various animals)

LLM Name	OPT 30B Erebus
Repository 🤗	https://huggingface.co/KoboldAI/OPT-30B-Erebus
Model Size	30b
Required VRAM	36 GB
Updated	2026-08-03
Maintainer	KoboldAI
Model Type	opt
Model Files	1.0 GB: 1-of-74 0.6 GB: 2-of-74 0.8 GB: 3-of-74 0.9 GB: 4-of-74 0.7 GB: 5-of-74 0.8 GB: 6-of-74 0.9 GB: 7-of-74 0.7 GB: 8-of-74 0.8 GB: 9-of-74 0.9 GB: 10-of-74 0.7 GB: 11-of-74 0.8 GB: 12-of-74 0.9 GB: 13-of-74 0.7 GB: 14-of-74 0.8 GB: 15-of-74 0.9 GB: 16-of-74 0.7 GB: 17-of-74 0.8 GB: 18-of-74 0.9 GB: 19-of-74 0.7 GB: 20-of-74 0.8 GB: 21-of-74 0.9 GB: 22-of-74 0.7 GB: 23-of-74 0.8 GB: 24-of-74 0.9 GB: 25-of-74 0.7 GB: 26-of-74 0.8 GB: 27-of-74 0.9 GB: 28-of-74 0.7 GB: 29-of-74 0.8 GB: 30-of-74 0.9 GB: 31-of-74 0.7 GB: 32-of-74 0.8 GB: 33-of-74 0.9 GB: 34-of-74 0.7 GB: 35-of-74 0.8 GB: 36-of-74 0.9 GB: 37-of-74 0.7 GB: 38-of-74 0.8 GB: 39-of-74 0.9 GB: 40-of-74 0.7 GB: 41-of-74 0.8 GB: 42-of-74 0.9 GB: 43-of-74 0.7 GB: 44-of-74 0.8 GB: 45-of-74
Supported Languages	en
Model Architecture	OPTForCausalLM
License	other
Context Length	2048
Model Max Length	2048
Transformers Version	4.25.1
Vocabulary Size	50265
Torch Data Type	float16
Activation Function	relu

Rank the OPT 30B Erebus 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 OPT 30B Erebus

Best Alternatives	Context / RAM	Downloads	Likes
Galpaca 30B MiniOrca	2K / 59.6 GB	48	1
Opt 30B	2K / 60.1 GB	14167	137
Galpaca 30B	2K / 60.8 GB	59	55
Opt Iml Max 30B	2K / 60.1 GB	1150	35
...alactica 30B Evol Instruct 70K	2K / 60.1 GB	57	23
Opt Iml 30B	2K / 60.1 GB	36	73
Galactica 30B	2K / 60.8 GB	71	40
Dalio Io 30B Cp1500	2K / 121.7 GB	11	0
Galactica 30B	2K / GB	19	1
...h 1e 5 Lr Handwritten Complete	2K / 121.7 GB	18	0

Note: green Score (e.g. "73.2") means that the model is better than KoboldAI/OPT-30B-Erebus.