AI Intel

A Reddit user warns that the RTX 5090 MSI's power cable is at risk of bending when running at 475-500W for inference or training.

A user reported that the GLM 5.2 UD IQ2_M model, despite low quantization, demonstrated 'capable as hell' performance on an RTX 5090 and RTX 3090 mul…

WorkClaw has launched a tool that provides collaborative, proactive AI coworkers directly within Slack.

OpenAIDevs issued a reminder that the Codex App, CLI, and SDK are compatible with any open source model, not exclusively OpenAI models.

Google's Nano Banana 2 and Nano Banana Pro models are now generally available via the Gemini Enterprise Agent Platform, API, and Google AI Studio, al…

Mistral Vibe has launched as an AI agent designed for long-horizon productivity and coding, integrating Work mode, Code mode, a CLI, and a new VS Cod…

vLLM has released version v0.23.0, featuring 408 commits from 200 contributors, though Minimax M3 support is not yet included.

Basedash has released new access control features, enabling companies to precisely manage who can access their data.

Slackbot's MCP Client now facilitates multiplayer collaboration across more than 20 applications directly within Slack.

Gemini Code Assist announced the general availability of Gemini 2.5 Pro and Flash models, adding agent mode and AI exclusion file configuration for V…