←
VentureBeat181,509 posts
VentureBeat avatar

VentureBeat

@VentureBeat

Obsessed with covering transformative technology.

◷ Joined July 2009▧ 38,999 Photos and videos
VentureBeat avatar
VentureBeat@VentureBeat·14h

'Essentially unusable': OpenAI launches GPT-6.2 Astra and Sol to negative reception from AI power users venturebeat.com/ai/essential…

VentureBeat avatar
VentureBeat@VentureBeat·19h

2026 made inference cheaper. 2027 will test where context lives. Every agent step needs working memory. When that memory is slow, GPUs sit idle and the same tokens get recomputed over and over. We're unpacking it Dec 10 in Menlo Park with @WEKA + @Adobe: bit.ly/4iSV7pR

Post attachment
VentureBeat avatar
VentureBeat@VentureBeat·Nov 17

Alibaba’s Qwen4-Max cuts AI task costs, but benchmarks show an uneven challenge to frontier rivals venturebeat.com/ai/alibabas-…

VentureBeat avatar
VentureBeat@VentureBeat·Nov 17

“There’s no gap left to measure.” CrowdStrike is moving away from breakout time as AI agents speed up attacks. Its executives haven’t agreed on a replacement. venturebeat.com/security/ai-…

0:19
VentureBeat avatar
VentureBeat@VentureBeat·Nov 16

Read the patch, find the fix, write the exploit, run it. Google found an attacker framework built to do all four with no human involved. What Google hasn’t shown: that the whole chain worked. venturebeat.com/ai/google-sa…

VentureBeat avatar
VentureBeat@VentureBeat·Nov 16

New VB AI Impact Series salon: The Context Layer: Where Inference Gets Won or Lost in 2027 Agents need working memory at every step. Where should it live? With Liran Zvibel (@WEKA) + Aanchal Gupta (@Adobe), hosted by Matt Marshall Thu Dec 10 · Menlo Park Request an invite: bit.ly/4f5zku2

Post attachment