OpenAI reports agents generated 55% of output tokens in its EU business accounts, but the measure does not show how many tasks or how much value they produced.
EU Business AI Is Mostly Agentic, Says OpenAI—But What’s Being Counted?
Browsing Category
Tech
736 posts
The Clockwork Close: A Look At “Nightmarket — Mercado De Medianoche” (FABLE/175)
A ThorstenMeyerAI.com field guide examines how Nightmarket’s clock-driven changes make a digital market feel as if it is closing.
Inside The World-Building For “Luciole — The Synchronous Fireflies Of Vallon Creux”
Room 155 of the FABLE/175 exhibition uses a simulated firefly field to show how local interactions can produce synchronized flashes.
Jev Can See Now: My Experiment With 30 Personal Pictures
A writer tested TypeSafe’s new Jev image features on 30 site images, reporting strong product-photo results and weaker fit for news images.
Portable Power Station Shopping For 2027: 15 Picks For Camping And Backup
A 15-model portable power station guide compares capacity, output, battery chemistry and portability for camping and home backup.
Tech Leaders Warn As Anthropic Sets Claude Cruelty Rules
A Yahoo Finance headline says Anthropic will ban cruelty toward Claude from Nov. 12, but the policy, date year and tech leaders’ warning lack context.
A Hands-On AI Coding Agent Project: My Retirement Countdown
Thorsten Meyer says an AI coding agent helped build a free retirement countdown app in five days. He directed the project and approved its releases.
FrontierCode: How Maintainers Respond To AI-Generated Pull Requests
Artificial Analysis tested 23 models on FrontierCode. The top model passed every blocking check in 57.7% of attempts.
Evaluating AI Configurations For News: Notes From My 71-Setup Test
A 71-configuration test favored GPT-6 Astra for article quality but GPT-6 Luna for production cost; two fallback models were removed.
Muse Wonder Full Joins The AI Frontier, With Muse Wonder Charging Like Before
Artificial Analysis scores Muse Wonder Full at 56, while API-available Muse Wonder scores 52 and costs $4.52 per benchmark task.
