SpaceXAI Unveils Grok 4.5, Promising Enhanced Efficiency and Lower Costs

AI Grok 4.5 SpaceXAI

Hızlı Bakış

SpaceXAI has launched Grok 4.5, a new AI model emphasizing "workhorse" capabilities in coding, office tasks, and research.
The company claims "twice greater token efficiency" compared to competitors, potentially lowering costs for users.
Elon Musk compared it to Anthropic's Opus model, noting Grok 4.5 is faster and cheaper.

Yapay zekâ özeti

Yazı boyutu

SpaceXAI has launched Grok 4.5, a new AI model emphasizing "workhorse" capabilities in coding, office tasks, and research. The company claims "twice greater token efficiency" compared to competitors, potentially lowering costs for users. Elon Musk compared it to Anthropic's Opus model, noting Grok 4.5 is faster and cheaper.

Tam haberi TechCrunch sitesinde oku

İlgili Konular

KonularGrok 4.5 SpaceXAI Elon Musk token efficiency LLM Opus

Bu haber ilk olarak şurada yayınlandı: TechCrunch.

İlgili Haberler

Bu konuda daha fazlaAI

Gelişiyor·2 sa önce

Meta Reportedly Developing Always-On AI Smart Glasses with Continuous Recording

Meta is reportedly working on prototype "super sensing" smart glasses that could continuously record audio and snap photos. Data metadata would be uploaded for AI queries, raising privacy concerns, especially as the LED recording indicator may remain off during "AI Feature" use.

The Verge

Acil·3 sa önce

Two High-Severity Linux Vulnerabilities Surface, One Exploitable by Untrusted VMs

Two critical Linux vulnerabilities, Januscape (CVE-2026-53359) and GhostLock (CVE-2026-43499), have been disclosed. Januscape allows untrusted virtual machines to gain root access to host machines, while GhostLock enables limited users to escalate privileges to root.

Ars Technica

Acil·3 sa önce

NHTSA Directs AV Developers to Stop Interfering with First Responders

The NHTSA has issued a directive to autonomous vehicle developers, demanding immediate solutions to prevent AVs from interfering with first responders and law enforcement. The agency cited a pattern of driverless cars blocking emergency scenes and failing to recognize safety conditions.

TechCrunch

Gelişiyor·4 sa önce

Google's SynthID Watermark Debunks Fake Mitch McConnell Image

Google's SynthID system successfully identified a widely circulated hoax image of Senator Mitch McConnell as AI-generated, thanks to its invisible watermark. The technology, integrated into AI models like Gemini, aims to combat deepfake hoaxes.

TechCrunch

Gelişiyor·5 sa önce

Samsung Offers $30 Credit for Upcoming Galaxy Z Fold Preorders

Samsung is offering a $30 store credit for customers who preorder the upcoming Galaxy Z Fold phones after registering their interest. The credit must be used at checkout for accessories or services like Samsung Care Plus, not on the phone itself. The Galaxy Unpacked event is scheduled for July 22nd, 2026.

The Verge

Gelişiyor·5 sa önce

General Intuition CEO: Embodied AI will follow foundation model pattern

General Intuition CEO Pim de Witte believes embodied AI will adopt a foundation model approach, similar to NLP. The company raised $320M at a $2.3B valuation for its model trained on video game data, capable of powering robots with minimal real-world data.

TechCrunch

Bu konuda daha fazlaAI