What are the hardware requirements for Opt 13B?

Opt 13B requires approximately 25.8 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 13B and how large is it?

Opt 13B is developed by facebook, a model with 13b parameters. The model is published as open weights on Hugging Face and indexed on LLM Explorer with full benchmark history.

How does Opt 13B perform on standard benchmarks?

Opt 13B has the following published scores: MMLU 24.9. Compare against reference models on this page or on the LLM Explorer leaderboards.

Opt 13B by facebook — VRAM 25.8GB, 2K context

Name: Opt 13B
Author: facebook

Opt 13B is an open-source language model by facebook. Features: 13b LLM, VRAM: 25.8GB, Context: 2K, License: other, LLM Explorer Score: 0.08, Arc: 39.9, HellaSwag: 71.2, MMLU: 24.9, GSM8K: 1.7.

Arxiv:2005.14165 Arxiv:2205.01068 En Jax Opt Pytorch Region:us Sharded Tf

Model Card on HF 🤗: https://huggingface.co/facebook/opt-13b

Opt 13B Parameters and Internals

Model Type

text generation, multimodal

Use Cases

Areas:

research, commercial applications

Applications:

text generation, evaluation of downstream tasks

Primary Use Cases:

prompting for evaluation, fine-tuning on downstream tasks

Limitations:

bias, limited diversity in generation, hallucination

Supported Languages

languages_and_proficiency ()

Training Details

Data Sources:

BookCorpus, CC-Stories, The Pile, Pushshift.io Reddit dataset, CCNewsV2

Data Volume:

180B tokens (~800GB of data)

Methodology:

Pre-trained using causal language modeling objective.

Context Length:

2048

Training Time:

~33 days

Hardware Used:

992 * 80GB A100 GPUs

Model Architecture:

Decoder-only Transformers

Responsible Ai Considerations

Fairness:

Model shows biased predictions as it is trained on unfiltered internet data.

Transparency:

Model card available in the official paper.

Input Output

Input Format:

Prompts

Accepted Modalities:

text

Output Format:

Generated text

Performance Tips:

Load model in half-precision for large models, use `generate` method.

LLM Name	Opt 13B
Repository 🤗	https://huggingface.co/facebook/opt-13b
Model Size	13b
Required VRAM	25.8 GB
Updated	2026-05-30
Maintainer	facebook
Model Type	opt
Model Files	10.0 GB: 1-of-3 9.9 GB: 2-of-3 5.9 GB: 3-of-3
Supported Languages	en
Model Architecture	OPTForCausalLM
License	other
Context Length	2048
Model Max Length	2048
Transformers Version	4.21.0.dev0
Tokenizer Class	GPT2Tokenizer
Beginning of Sentence Token	</s>
End of Sentence Token	</s>
Unk Token	</s>
Vocabulary Size	50272
Torch Data Type	float16
Activation Function	relu
Errors	replace

Rank the Opt 13B 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 13B

Best Alternatives	Context / RAM	Downloads	Likes
Emoopt 13B	2K / 25.7 GB	8	0
OPT 13B Nerybus Mix	2K / 26.4 GB	171	36
OPT 13B Erebus	2K / 25.7 GB	413	261
OPT 13B Nerys V2	2K / 25.7 GB	110	12
Basic Facebook 13B	2K / 25.8 GB	15	1
Opt 13B	2K / 26.3 GB	7	1
OPT 13B Nerybus Mix 4bit 128g	2K / 7.3 GB	58	6
OPT 13B Erebus 4bit 128g	2K / 7.3 GB	8	17

Note: green Score (e.g. "73.2") means that the model is better than facebook/opt-13b.