
Zhipu’s GLM 5.2 closes on US leaders, leaving Anthropic and OpenAI “held back”
The model benchmark shift makes “intelligence per dollar” the real battleground, and open source suddenly matters.

Pulling the latest in this topic…
AI, robotics, computing, technical advancement
978 briefings briefings in Technology

The model benchmark shift makes “intelligence per dollar” the real battleground, and open source suddenly matters.

The software rally lifted many names, but Oracle missed out because OpenAI success also runs through its cloud infrastructure.

After Papermark accused Corgi of stealing software, Corgi says it did not. Here’s what that means for code trust in startups.

A 60-year-old headache case turned from oncology suspicion to parasitic reality, highlighting why diagnostic assumptions can mislead.

New rules for Gavin Newsom's SB 576 force streamers to match ad loudness to content, starting July 1.

Meta is building Arena, and Zuckerberg is pushing the app toward existing prediction market players like Polymarket and Kalshi.

The automaker says it believed it could “swap in AI” without losing product quality, then had to fix the process.

A phased rollout starts for a small trusted group, with full general availability aimed for coming weeks.

A niche manufacturing step now functions like a choke point for AI compute, forcing new risk management decisions.

A 31 August 2025 attack tied to Russian hackers stalled Jaguar Land Rover production and hit the British economy hard.

A frontier model goes live under a named-partner list, signaling a shift from voluntary reviews to managed rollout.

Custom inference chips are multiplying, and decision-makers now have to manage the build-versus-buy power shift.