techimo.bsky.social
Tech news. Every day. No noise.
👤 Joined on 23 January 2026
Followers
Following
Total Posts
@techimo.bsky.socialNvidia boosted Kimi K3 model inference speed by 1.68x to 423 tokens/s/user on GB300 NVL72 hardware, thanks to vLLM optimization by CentML engineers. They're also optimizing models like DeepSeek V4.1.
Show PostiPhone Duo launch day (Oct 23rd) will have *very* limited stock. Apple advises against walk-ins. Production delays & unprecedented demand led to extended pre-order delivery times (Jan next year). China price: 15,999 yuan.
Show PostPwnEscape vulnerability allows escape from Linux KVM! Client-controlled metadata overwrites IPv4 header after kernel check, leading to host memory leak & full escape. Patched. Developers warned about "deferred checksum" risks. CVE-2026-99999.
Show PostMeta's terms of service update for AI training data sparked discussion. Initially, it seemed to allow using business account web content for AI training. This statement has since been removed. Current terms limit data collection to advertising & commercial products. — Sina News, The Verge
Show PostGerman CDU party hacked! 😳 ~6.5k contacts got malicious emails with phishing links disguised as event invites. Stay safe, click carefully! #cybersecurity #hack #CDU
Show Post