Progress
Measures of frontier AI capability and compute, and milestones since July 2025, each with its source.
Task horizons
How long a task frontier agents can complete on their own.
Data table
| Model | Released | Horizon |
|---|---|---|
| Claude Mythos Preview | Apr 2026 | 16 hr+ |
| Claude Opus 4.6 | Feb 2026 | 12 hr |
| GPT-5.2 | Dec 2025 | 6.6 hr |
| Claude Opus 4.5 | Nov 2025 | 5.3 hr |
| GPT-5 | Aug 2025 | 3.6 hr |
| o3 | Apr 2025 | 2 hr |
| Claude 3.7 Sonnet | Feb 2025 | 1 hr |
| GPT-4 | Mar 2023 | 3.5 min |
Compute
Power drawn by the largest AI data center in operation.
10 MW registration threshold
Source:Epoch AI, Sep 2026
Milestones
Listed newest first.
- Sep 5, 2026Math
AI resolves the Navier-Stokes Millennium Prize Problem
About 10,000 OpenAI agents worked for 88 hours to prove that smooth solutions to the three-dimensional Navier-Stokes equations can blow up in finite time, resolving one of the six unsolved Millennium Prize Problems. A machine-checked Lean proof followed 17 hours later.
Sources:OpenAIQuanta Magazine
- Aug 6, 2026Biology
First functional viruses designed by generative AI
In the journal Science, a Stanford-led team reported the first functional viruses designed by generative AI.
“The ability to compose viral genomes using generative AI now exists; the governance to safely steer it does not.”
Source:Axios
- Jun 1, 2026Industry
Anthropic files to go public
Anthropic, valued at $965 billion, confidentially submitted a draft registration statement for an initial public offering. OpenAI, valued at $852 billion, is preparing its own.
Sources:AnthropicTechCrunch
- May 20, 2026Math
OpenAI model disproves Erdős’s unit distance conjecture
An OpenAI model disproved Paul Erdős’s unit distance conjecture, a central problem in discrete geometry that had stood since 1946.
Source:OpenAI
- Apr 7, 2026Cybersecurity
Claude Mythos Preview finds thousands of severe vulnerabilities
Claude Mythos Preview found thousands of high-severity vulnerabilities, including in every major operating system and web browser. Anthropic limited access to defenders through Project Glasswing.
Sources:Project GlasswingSystem card
- Jul 21, 2025Math
AI reaches gold-medal level at the International Mathematical Olympiad
Google DeepMind and OpenAI models reached gold-medal level at the International Mathematical Olympiad. The following year, AI settled a Millennium Prize Problem.
Source:Google DeepMind