0.731
Based on primary energy power · Type I = 1

AccelerationFour-dimensional view2 years agoAI Agents
An open-source large model reaching the frontier at low training cost.
Art style · New Chinese ink wash with cool light
Each sparkline shows the documented history of a metric on a logarithmic time axis. The dot marks the selected moment.
0.731
Based on primary energy power · Type I = 1
8.23Bpeople
Comparable to today
2×10¹³W
Comparable to today
7.9×10¹⁰floating-point operations per second per dollar (FLOP/s per $, nominal dollars)
Comparable to today
0.03 seconds
Comparable to today
39,897km/h
Comparable to today
The eight closest milestones. Select a title to jump there.

Models think before answering, making test-time compute a new scaling dimension. Source
Hopfield and Hinton are honored for foundational work on machine learning with artificial neural networks. Source
Hassabis and Jumper win for protein structure prediction, sharing the prize with Baker. Source

On Starship's fifth flight test, the Super Heavy booster returns to the launch tower and is caught by its mechanical arms. Source

Anthropic introduces "computer use": the model can see the screen, move the cursor, click and type. Source

An open-source large model reaching the frontier at low training cost. Source

A frontier-level reasoning model released with open weights under the MIT license. Source

SoftBank, OpenAI, Oracle and MGX pledge $500 billion over four years to build AI data centers in the US. Source
Acquired121 / 121 items
All documented capabilities have appeared. What follows is forecast.
Each era illustrated in its own style.
Every timeline entry cites its source. Metrics interpolate logarithmically between sourced data points. Forecasts include institutional and community views, and attributed site scenarios; years are not guaranteed realization dates. Illustrations are AI-generated in each era's style.