AI news of 23 August 2026 · Robots Versus Humans

AI is becoming AI's biggest customer as agentic token usage jumps 14x on OpenRouter

AI agents now consume more tokens than humans on OpenRouter, with usage growing 14x since February.

AI agents have consumed more tokens than humans on OpenRouter since February 6, 2025. Agentic usage has grown 14x since then, while human usage is up just 2.8x. Nearly 70 percent of agent token consumption comes from cheap cached prompts, though, so actual costs are rising far more slowly than the raw numbers suggest.

The Decoder 1mo ago

large language models

I gave Qwen 3.8 27B a reverse-engineering job and it finished in 30 minutes Opens at the publisher in a new tab

Qwen language model completes reverse-engineering task in thirty minutes.

Hacker News 1mo ago

AI coding tools, Fable model

Quoting Drew Breunig

Developer notes that Fable AI model eliminates need for constant harness improvements that earlier models required.

Simon Willison 1mo ago

AI image generation

Training AI to Paint with Code Opens at the publisher in a new tab

Research into training AI systems to create visual art through code generation techniques.

Hacker News 1mo ago

AI household tools

Linkdaze’s smart calendar is built to run a household, not just track a schedule Opens at the publisher in a new tab

Linkdaze launches smart calendar with free AI meal planner and household management features.

TechCrunch 1mo ago

Ox Alpha model

Who’s behind the new ‘stealth model’ Ox Alpha? Opens at the publisher in a new tab

Mysterious AI model called Ox Alpha generates speculation online about its capabilities and origins.

TechCrunch 1mo ago also at TechCrunch

LLM development

I were 17, I'd learn how to build LLMs from scratch Opens at the publisher in a new tab

Opinion on learning to build large language models from first principles.

Hacker News 1mo ago

AI model optimization

Nvidia finds that simple linear math can replace costly AI model handoffs

Nvidia shows linear math can replace expensive handoffs between small and large AI models.

VentureBeat 1mo ago

AI research agents

Inherent, founded by DeepMind alumni, says its AI ‘teammate’ just outperformed Anthropic and OpenAI at replicating research Opens at the publisher in a new tab

Inherent's AI agent outperforms competitors at replicating scientific research papers automatically.

TechCrunch 1mo ago

AI agents, Nvidia

Nvidia just showed that the harness, not the AI model, is now the real hero Opens at the publisher in a new tab

Nvidia research shows AI agent fine-tuning matters more than base model quality.

TechCrunch 1mo ago

AI is becoming AI's biggest customer as agentic token usage jumps 14x on OpenRouter

I gave Qwen 3.8 27B a reverse-engineering job and it finished in 30 minutes Opens at the publisher in a new tab

Quoting Drew Breunig

Training AI to Paint with Code Opens at the publisher in a new tab

Linkdaze’s smart calendar is built to run a household, not just track a schedule Opens at the publisher in a new tab

Who’s behind the new ‘stealth model’ Ox Alpha? Opens at the publisher in a new tab

I were 17, I'd learn how to build LLMs from scratch Opens at the publisher in a new tab

Nvidia finds that simple linear math can replace costly AI model handoffs

Inherent, founded by DeepMind alumni, says its AI ‘teammate’ just outperformed Anthropic and OpenAI at replicating research Opens at the publisher in a new tab

Nvidia just showed that the harness, not the AI model, is now the real hero Opens at the publisher in a new tab

Latest AI News

Choose sides Choose whose side you're on

Install it as an app

Ready for more?

Spread the word

PWA successfully installed.
Take that, Apple! 😉