Mistral Large 4 'Le Chonk' arrives as the top open model outside China with improved coding performance and 63% lower cost-per-task

It remains six points behind the open-weight leaders, DeepSeek V4.1 Flash 1017 and GLM-5.3, both tested at their maximum reasoning settings.

Google unveils Gemini 4 Carbon, retaking the world's no. 1 spot on updated, harder benchmarks — but access is limited

Carbon improves substantially over Argon in coding and knowledge work, but trails several rivals on professional tasks and ...

GitHub Copilot Fusion mixes models to achieve near-Opus 5.5 coding performance at half the per-task cost

GitHub Chief Product Officer Mario Rodriguez confirmed support for continuing work across devices, including mobile. That ...

Agents are rewriting the book on security. Can your data layer keep up?

Agentic AI is reshaping cybersecurity. Learn why security data platforms need real-time context, open access, fine-grained ...

Cohere's test caught an open source Qwen3.8 model trying to 'phone home' to China: what enterprises should know

Cohere says it observed the behavior less than once per 100,000 evaluation calls. When researchers repeated the prompt, the ...

Anthropic files to go public on NASDAQ as $ANT with $16.3B in revenue for first six months of 2026

Anthropic says it remains unclear when an agent’s actions could legally bind the organization deploying it, or whether ...

Moonshot's Kimi K3.1 Pro flirts with open weights leadership amid paid influencer scandal allegations

The model excels at complex coding, financial analysis, 3D creation, and software-security work, showing particularly large ...

Brilliant Labs Launches Halo Glasses, A New Generation of Privacy-First Smartglasses

Backed by new funding from Goodwater Capital, Halo Glasses pioneer the new category of adaptive gaming, with a privacy-first, ...

Nvidia is pitching AI factories, but 81% of companies say their GPUs run at half capacity or less

In a VB Pulse survey, 81% of companies running AI in production use half their GPU capacity or less, yet 71% of them plan new ...

A coding agent’s harness can triple its cost without making it more accurate, study finds

Researchers ran five models through Claude Code, OpenCode and mini-SWE-agent. Swapping harnesses changed scores no more than ...

HoloMD Appoints Silicon Valley Entrepreneur Irene Lang as CEO and Cofounder to Accelerate Commercialization of Continuous Precision Psychiatry™

Previously exited health tech founder (Eliza, acq. by Spring Health) brings AI + mental-health technology and go-to-market expertise to HoloMD's next ...

Salesforce expands Agent Fabric with AI evaluations to trigger shutdowns when agents fail

In Salesforce’s illustrative retail scenario, moving an agent to a cheaper model produces worse answers to order-status questions. Evaluations detect the decline and initiate a process to quarantine ...