Timeline of AI and language models

Show older timeline slides

Download source (PDF)


Time between releases of OpenAI’s GPT models

Model Months since last release
GPT-4
Mar/2023
14m
14 months
GPT-3 2022 text-davinci-002
Jan/2022
20m
20 months
GPT-3
May/2020
15m
15 months
GPT-2
Feb/2019
8m
8 months
GPT-1
Jun/2018
Baseline

Full AI timeline

Showing language and multimodal models only (less focus on text-to-image generation models etc). Showing Western models mostly (less focus on China, South Korea). Showing selected major milestones in AI development. Yes, the timeline is ordered by year descending, then month ascending, for my own amusement.

Next…

TBA

Inflection (model)

Inflection AI

TBA

TBA

70B (model)

Stability AI

TBA: Emad Feb/2023: ‘have the new language and code ones training [now]… Should outperform [Meta AI’s latest 65B model] llama in Lm side at least. Doubt anyone needs more than 70bn parameters’

TBA

Project Gemini 1T (model)

Google/DeepMind

[bold_timeline_item_button title=”Expand” style=”” shape=”” color=”” size=”inline” url=”#” el_class=”bold_timeline_group_button”]

2023

January

Pile of law (dataset)

Stanford

January

FLAME 60M (model)

Microsoft

February

Multimodal-CoT 738M (model)

Amazon

February

Toolformer 6.7B (+Atlas 11B+NLLB 54B) (model)

Meta AI

February

Luminous-Supreme-Control 70B (model)

Aleph Alpha

February

Palmyra 20B (model)

Writer

February

MOSS 20B (chatbot model)

Fudan University

February

LLaMA-65B (model)

Meta AI

February

Kosmos-1 1.6B (model)

Microsoft

March

GPT-NeoX-Chat-Base-20B (model)

Together

March

Alpaca 7B (model)

Stanford

March

GPT-4 1T (model)

OpenAI

March

Med-PaLM 2 (model)

Google

March

CoLT5 5.2B (model)

Google

March

Cerebras-GPT 13B (model)

Cerebras

March

BloombergGPT 50B (model)

Bloomberg

[bold_timeline_item_button title=”Expand” style=”” shape=”” color=”” size=”inline” url=”#” el_class=”bold_timeline_group_button”]

2022

January

GPT-3.5 – text-davinci-002 (model)

OpenAI

This is a new model with data ending Jun/2021, context window of 4,000 tokens
Read the OpenAI blog

January

CM3 13B (model)

Meta AI

February

GPT-NeoX-20B (model)

EleutherAI

February

xlarge 52.4B (model)

Cohere

March

Chinchilla 70B (model)

DeepMind

March

BLOOM – tr11-176B-ml (model)

BigScience

March

SeeKeR 2.7B (chatbot model)

Meta AI

April

PaLM 540B (model)

Google Inc

April

PaLM-Coder 540B (model)

Google Inc

April

CodeGen 16B (model)

Salesforce

April

VLM-4 (model)

LightOn

April

Luminous 200B (model)

Aleph Alpha

April

mGPT 13B (model)

Sber

April

NOOR 10B (model)

TII

April

InCoder 6.7B (model)

Meta AI

April

Flamingo (Chinchilla 70B + 10B visual model)

DeepMind

May

OPT-175B (model)

Meta AI

May

LaMDA 2 137B (chatbot model)

Google AI

May

Gato (Cat) 1.18B (general model)

DeepMind

May

UL2 20B (model)

Google Research

May

Diffusion-LM 300M (model)

Stanford

June

Perceiver AR (model)

DeepMind

June

Unified-IO 2.8B (model)

Allen AI

June

YaLM 100B (model)

Yandex

June

GODEL-XL 2.7B (chatbot model)

Microsoft/Columbia

June

Minerva 540B (model)

Google Research

July

No Language Left Behind (NLLB) 54.5B/MoE (model)

Meta AI

July

PanGu-Coder 2.6B (model)

Huawei

July

monorepo-Transformer 0.5B (model)

Google Brain

July

FIM 6.9B (model)

OpenAI

August

AlexaTM 20B (model)

Amazon Alexa AI

August

GLM-130B (model)

Tsinghua & Zhipu

August

BlenderBot 3 (chatbot model)

Meta AI

August

Atlas 11B (model)

Meta AI

August

Z-Code++ 710M (model)

Microsoft

September

PaLI 17B (visual model)

Google

September

Sparrow 70B (chatbot as fine-tuned Chinchilla)

DeepMind

September

CodeGeeX 13B (code)

Tsinghua

September

WeLM 10B (model)

WeChat

October

VIMA 200M (general model)

NVIDIA

October

U-PaLM 540B (model)

Google

October

Flan-PaLM 540B (model)

Google

October

Flan-T5 11B (model)

Google

October

PACT (model)

Microsoft

November

BLOOMZ 176B & mT0 13B (models)

BigScience

November

SED 420M (diffusion text model)

DeepMind

November

Galactica 120B (model)

Meta AI

November

RWKV-4 7B & 14B (RNN model)

EleutherAI

November

The Stack (code dataset)

ServiceNow

November

GPT-3.5 – text-davinci-003 (model)

OpenAI

November

GPT-JT 6B (model)

Together

November

ChatGPT (model)

OpenAI

December

RT-1 35M (general model)

Google

December

ERNIE-Code 560M (code)

Baidu

December

RL-CAI 52B (model)

Anthropic

December

OPT-IML 175B (model)

Meta AI

December

Med-PaLM 540B (model)

Google & DeepMind

[bold_timeline_item_button title=”Expand” style=”” shape=”” color=”” size=”inline” url=”#” el_class=”bold_timeline_group_button”]

2021

January

The Pile v1 (dataset)

EleutherAI

March

Wudao 1.0 (model)

BAAI

June

GPT-J-6B (model)

EleutherAI

June

LaMDA 137B (chatbot model)

Google

June

Wudao 2.0 (model)

BAAI

June

M6 1T – MultiModality-to-MultiModality Multitask Mega-transformer (sparse model)

Alibaba Dharma Academy

July

BlenderBot 2.0 9.4B (chatbot model)

Facebook

July

Codex 12B (model)

OpenAI

August

Jurassic-1 178B (model)

AI21

September

PLATO-XL 11B (chatbot model)

Baidu

September

Macaw 11B (Q&A model)

AI2 (Allen AI)

October

Megatron-Turing NLG 530B (model)

NVIDIA + Microsoft

October

Yuan 1.0 245B (model)

Inspur AI (China)

November

M6 10T – MultiModality-to-MultiModality Multitask Mega-transformer (sparse model)

Alibaba Dharma Academy

November

Cedille 6B (model, French, based on GPT-J 6B)

Coteries

November

BERT 480B & 200B (model)

Google

December

52B (model)

Anthropic

December

GLaM 1.1T (model)

Google inc

December

Gopher 280B (model)

Google AI

December

Fairseq-13B (model)

Meta AI

December

ERNIE 3.0 Titan 260B (model)

Baidu

[bold_timeline_item_button title=”Expand” style=”” shape=”” color=”” size=”inline” url=”#” el_class=”bold_timeline_group_button”]

2020

January

Meena 2.6B (chatbot model)

Google

April

BlenderBot 1.0 (chatbot model)

Facebook

May

GPT-3 175B (model)

OpenAI

September

GPT-3 writes a newspaper column

The Guardian/OpenAI

[bold_timeline_item_button title=”Expand” style=”” shape=”” color=”” size=”inline” url=”#” el_class=”bold_timeline_group_button”]

2019

February

GPT-2 1.5B (model)

OpenAI

October

BERT used for search

Google

[bold_timeline_item_button title=”Expand” style=”” shape=”” color=”” size=”inline” url=”#” el_class=”bold_timeline_group_button”]

2018

June

GPT-1 117M (model)

OpenAI

[bold_timeline_item_button title=”Expand” style=”” shape=”” color=”” size=”inline” url=”#” el_class=”bold_timeline_group_button”]

2017

August

Transformer (architecture)

Google

[bold_timeline_item_button title=”Expand” style=”” shape=”” color=”” size=”inline” url=”#” el_class=”bold_timeline_group_button”]

2011

February

Watson (system)

IBM

Appeared on Jeopardy! against champions Brad Rutter and Ken Jennings, winning the first place prize of $1m

[bold_timeline_item_button title=”Expand” style=”” shape=”” color=”” size=”inline” url=”#” el_class=”bold_timeline_group_button”]

1966

ELIZA (chatbot)

MIT

[bold_timeline_item_button title=”Expand” style=”” shape=”” color=”” size=”inline” url=”#” el_class=”bold_timeline_group_button”]

1950

Turing test (paper)

[bold_timeline_item_button title=”Expand” style=”” shape=”” color=”” size=”inline” url=”#” el_class=”bold_timeline_group_button”]

1947

Turing lecture

First public lecture (London, 1947) to mention computer intelligence. Turing said: ‘What we want is a machine that can learn from experience… the possibility of letting the machine alter its own instructions provides the mechanism for this.’ A few months later, he introduced many of the central concepts of AI in an unpublished paper: Intelligent Machinery. Britannica

[bold_timeline_item_button title=”Expand” style=”” shape=”” color=”” size=”inline” url=”#” el_class=”bold_timeline_group_button”]

Summary of current models


Summary of current models: View the full data (Google sheets)
Download PDF version


Get The Memo

by Dr Alan D. Thompson · Be inside the lightning-fast AI revolution.
Thousands of paid subscribers. Readers from Microsoft, Tesla, Google AI...
Artificial intelligence that matters, as it happens, in plain English.
Get The Memo.

Dr Alan D. Thompson is an AI expert and consultant, advising Fortune 500s and governments on post-2020 large language models. His work on artificial intelligence has been featured at NYU, with Microsoft AI and Google AI teams, at the University of Oxford’s 2021 debate on AI Ethics, and in the Leta AI (GPT-3) experiments viewed more than 2.5 million times. A contributor to the fields of human intelligence and peak performance, he has held positions as chairman for Mensa International, consultant to GE and Warner Bros, and memberships with the IEEE and IET. He is open to consulting and advisory on major AI projects with intergovernmental organizations and enterprise.

This page last updated: 15/Mar/2023. https://lifearchitect.ai/timeline/