OPT

Name: Open Pre-trained Transformer
Author: Facebook AI Research Lab

Foundation model

Open Pre-trained Transformer

Release date: 2022-05-02Updated: 2023-08-16Views: 1,002

Live demoGitHub Hugging Face Compare

Parameters

175B

Context length

Multilingual

No data

Reasoning ability

No data

Open Pre-trained Transformer is an AI model published by Facebook AI Research Lab, released on 2022-05-02, for Foundation model, with 175B parameters, and 2K context length, requiring about 85GB storage, with a 25.20 score on MMLU.

Data sourced primarily from official releases (GitHub, Hugging Face, papers), then benchmark leaderboards, then third-party evaluators. Learn about our data methodology

OPT

Model basics

Reasoning traces

No data

Thinking modes

Thinking modes not supported

Context length

2K tokens

Max output length

No data

Model type

Foundation model

Modality (in / out)

No data

Release date

2022-05-02

Model file size

85GB

MoE architecture

Total params / Active params

175B / Not applicable

Knowledge cutoff

No data

OPT

Open source & experience

Code license

MIT License

Weights license

Open source, non-commercial

GitHub repo

github.com/facebookresearch/metaseq/tree/main/p…

Hugging Face

huggingface.co/docs/transformers/model_doc/opt

Live demo

N/A

OPT

Official resources

Paper

DataLearnerAI blog

N/A

OPT

API details

API speed

No data

No public API pricing yet.

OPT

Benchmark Results

OPT currently shows benchmark results led by AGIEval (26 / 28, score 24.20), C-Eval (47 / 48, score 25), MMLU (122 / 124, score 25.20). This page also consolidates core specs, context limits, and API pricing so you can evaluate the model from benchmark results and deployment constraints together.