Back to News

Over 30 AI models have been trained at 10^25 FLOP or greater

Hacker News - AI
Jul 16, 2025 15:19
Jimmc414
1 views
hackernewsaidiscussion

Summary

Over 30 AI models have now been trained using at least 10^25 floating point operations (FLOP), marking a significant increase in the scale of AI training. This rapid growth in computational resources highlights accelerating progress in AI capabilities, but also raises concerns about resource concentration and the environmental impact of large-scale training.

Article URL: https://epoch.ai/data-insights/models-over-1e25-flop Comments URL: https://news.ycombinator.com/item?id=44583247 Points: 1 # Comments: 0
Read Full ArticleMore News

Related Articles

Grok 4 vs Grok 3: Key Performance Differences Explained

Analytics InsightJul 16

The article compares Grok 4 and Grok 3, highlighting that Grok 4 offers significant improvements in reasoning, language understanding, and coding abilities over its predecessor. These advancements position Grok 4 as a more competitive AI model, suggesting accelerated progress and heightened competition in the AI landscape.

Missed Early Binance Coin (BNB) Rally? Experts Say Ruvi AI (RUVI) Is Your Best Shot At Over 13800% ROI This Year

Analytics InsightJul 16

Experts suggest that Ruvi AI (RUVI), an emerging AI-focused cryptocurrency, could offer investors significant returns, potentially exceeding 13,800% this year, similar to the early rally seen with Binance Coin (BNB). This highlights growing investor interest in AI-integrated blockchain projects and suggests that AI-driven tokens may play an increasingly prominent role in the crypto market.

Tim O'Reilly -Where Is AI on the Enshittification Curve?

Hacker News - AIJul 16

Tim O’Reilly examines how AI technologies may be following the "enshittification curve," a process where platforms degrade in quality as they prioritize profits over user value. He warns that without careful stewardship, AI could repeat the mistakes of earlier tech platforms, undermining trust and long-term utility. This highlights the need for responsible development to ensure AI remains beneficial to users and society.