gpt4all.io Open in urlscan Pro
2606:4700:20::ac43:47a9  Public Scan

Form analysis
0 forms found in the DOM

Text Content

You need to enable JavaScript to run this app.
GPT4AllGPT4All


GPT4ALL

A free-to-use, locally running, privacy-aware chatbot. No GPU or internet
required.

Real-time inference latency on an M1 Mac

DOWNLOAD DESKTOP CHAT CLIENT

Windows InstallerOSX InstallerUbuntu Installer


PLEASE VIEW ON DESKTOP TO DOWNLOAD


GPT4ALL'S CAPABILITIES

Explore what GPT4All can do. On your own hardware.


ANSWER QUESTIONS ABOUT THE WORLD

Ask GPT4All about anything.




PERSONAL WRITING ASSISTANT

Write emails, documents, creative stories, poems, songs and plays.




UNDERSTAND DOCUMENTS

Provide your own text documents and receive summaries and answers about their
contents.




WRITE CODE

Get guidance on easy coding tasks. Code capabilities are under improvement.




INSTALLATION INSTRUCTIONS

WindowsMacOSUbuntu


WINDOWS INSTALLER

After download and installation you should be able to find the application in
the directory you specified in the installer. You will find a desktop icon for
GPT4All after installation.

NOTE: On Windows, the installer might show a security complaint. This is being
addressed as we're actively setting up cert sign for Windows.





PERFORMANCE BENCHMARKS

ModelBoolQPIQAHellaSwagWinoGrandeARC-eARC-cOBQAAvgGPT4All-J 6B
v1.073.474.863.464.754.93640.258.2GPT4All-J
v1.1-breezy7475.163.263.655.434.938.457.8GPT4All-J
v1.2-jazzy74.874.963.663.856.635.34158.6GPT4All-J
v1.3-groovy73.674.363.863.557.73538.858.1GPT4All-J Lora
6B68.675.866.263.556.435.740.258.1GPT4All LLaMa Lora
7B73.177.672.167.851.140.440.260.3GPT4All 13B
snoozy83.379.27571.360.944.243.465.3GPT4All
Falcon77.679.874.970.167.943.442.665.2Nous-Hermes79.578.98071.974.250.946.468.8Nous-Hermes283.980.780.171.375.752.146.270.0Nous-Puffin81.580.780.472.577.650.745.669.9Dolly
6B68.877.367.663.962.938.741.260.1Dolly 12B56.775.47162.264.638.540.458.4Alpaca
7B73.977.273.966.159.843.343.462.5Alpaca Lora
7B74.379.37468.856.643.942.662.8GPT-J 6.7B65.476.266.264.162.236.638.258.4LLama
7B73.177.47366.952.541.442.461.0LLama 13B68.579.176.270.16044.642.263.0Pythia
6.7B63.576.36461.161.335.237.256.9Pythia
12B67.776.667.363.863.934.83858.9Fastchat
T581.564.646.361.849.333.339.453.7Fastchat Vicuña
7B76.677.270.767.353.541.240.861.0Fastchat Vicuña
13B81.576.873.366.757.442.743.663.1StableVicuña
RLHF82.378.674.170.96143.544.465.0StableLM
Tuned62.571.253.654.852.431.133.451.3StableLM
Base60.167.441.250.144.9273246.1Koala 13B76.577.972.668.854.34142.862.0Open
Assistant Pythia 12B67.97868.16564.240.443.261.0Mosaic
MPT7B74.879.376.368.67042.242.664.8Mosaic
mpt-instruct74.380.477.267.872.244.64365.6Mosaic
mpt-chat77.178.274.567.569.443.344.264.9Wizard
7B78.477.269.966.556.840.542.661.7Wizard 7B
Uncensored77.774.26865.253.538.741.659.8Wizard 13B
Uncensored78.475.572.169.557.540.44462.5GPT4-x-Vicuna-13b81.37575.26558.743.943.663.2Falcon
7b73.680.776.367.37143.344.465.2Falcon 7b
instruct70.978.669.866.767.942.741.262.5text-davinci-00388.183.883.475.883.963.95175.7


GPT4ALL ECOSYSTEM

Explore the GPT4All open source ecosystem


GPT4ALL TRAINING

Train your own GPT4All models.


GPT4ALL DOCUMENTATION

Integrate a locally running LLM into any codebase.


GPT4ALL CHAT

A multi-platform chat interface for running local LLMs


GPT4ALL IN PYTHON

Python Bindings to GPT4All


GPT4ALL DATALAKE

An open-source datalake for donated GPT4All interaction data.


HOW GPT4ALL WORKS

GPT4All is an ecosystem to train and deploy powerful and customized large
language models that run locally on consumer grade CPUs.

The goal is simple - be the best instruction tuned assistant-style language
model that any person or enterprise can freely use, distribute and build on.

A GPT4All model is a 3GB - 8GB file that you can download and plug into the
GPT4All open-source ecosystem software. Nomic AI supports and maintains this
software ecosystem to enforce quality and security alongside spearheading the
effort to allow any person or enterprise to easily train and deploy their own
on-edge large language models.


MODEL EXPLORER

wizardlm-13b-v1.1-superhot-8k.ggmlv3.q4_0.bin

WIZARDLM-13B-V1.1-SUPERHOT-8K.GGMLV3.Q4_0.BIN

Best overall model

 * Instruction based
 * Gives very long responses
 * Finetuned with only 1k of high-quality data
 * Trained by Microsoft and Peking University
 * Cannot be used commercially


GPT4ALL DATASETS

To train a powerful instruction-tuned assistant on your own data, you need to
curate high-quality training and instruction-tuning datasets. Nomic AI has built
a platform called Atlas to make manipulating and curating LLM training data
easy.

You can find the latest open-source, Atlas-curated GPT4All dataset on
Huggingface. Make sure to use the latest data version.


GPT4ALL OPEN SOURCE DATALAKE

Data is one the most important ingredients to successfully building a powerful,
general purpose large language model. The GPT4All community has built the
GPT4All Open Source datalake as a staging ground for contributing instruction
and assistant tuning data for future GPT4All model trains. It allows anyone to
contribute to the democratic process of training a large language model.

All data contributions to the GPT4All Datalake will be open-sourced in their raw
and Atlas-curated form. You can learn more details about the datalake on Github.
You can contribute by using the GPT4All Chat client and 'opting-in' to share
your data on start-up. By default, the chat client will not let any conversation
history leave your computer.

Explore a recent snapshot of the GPT4All Datalake in Atlas below.

Atlas Map must be viewed on a larger display