DigestAI news desk

Cut through the AI noise.

Research6 min read

Epoch AI report finds AI costs drop 13x yearly, faster than Moore’s Law

A report from AI research firm Epoch AI suggests the price of AI has fallen by thousands of times in recent years, outpacing even Moore’s Law. The firm estimates costs drop under 50% every quarter and 13 times per year, faster than lithium batteries or DNA sequencing. For example, OpenAI’s GPT-o3 cost $0.30 per question on a benchmark in 2025, while GPT-5.6 Luna scored the same result for…

1 source

Key points

  • Epoch AI estimates AI costs drop **13x yearly**, faster than lithium batteries or DNA sequencing
  • OpenAI’s GPT-5.6 Luna costs **$0.0004 per question** on a benchmark vs. **$0.30** for GPT-o3 in 2025
  • Cost reductions slow to **32% per quarter** in 2026, possibly due to rising hardware/energy prices

The report highlights uneven progress across tasks: math benchmarks saw stagnation until mid-2026, while chess puzzles improved steadily. Epoch AI notes declining cost-reduction rates—down from 66% per quarter to 32%—possibly due to rising energy and hardware costs. Critics warn of ‘benchmaxxing,’ where models are optimized for specific tests rather than real-world use. The findings raise questions for labs like OpenAI and Anthropic, as cheaper alternatives may erode loyalty and revenue models.

The story so far

3 episodes →
  1. Epoch AI report finds AI costs drop 13x yearly, faster than Moore’s Lawthis story
Full story from Tom's Hardware · by Jon MartindaleOpen source ↗

The price of AI is crashing faster than the rate of Moore's Law, report suggests

Tom's Hardware · 30 September 2026

Loading the full article…

This text was published by Tom's Hardware and written by Jon Martindale. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗

Topics · follow one to build your own front page

The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.

Comments

via GitHub Discussions

More in Research

All →

Related stories