AI Intel

The SWE-rebench Leaderboard has been significantly updated to include recent models like GPT-5.5, Opus 4.7, and Kimi K2.6, reflecting performance on…

AI-generated CUDA kernels, including some from top-ranked models, have been found to silently break deep learning training and inference processes wh…

YouTube announced plans to automatically label videos that have been generated by AI, aiming to improve transparency for viewers and creators.

A developer successfully demonstrated a 260K-parameter LLM running within an 18-year-old RTOS, which itself was running on a JavaScript-emulated 90s…

Anthropic announced the acquisition of StainlessAPI, the SDK and MCP server platform that has underpinned all Anthropic SDKs since the early days of…

GoogleAI announced new upgrades to the Gemini App, introducing Gemini Spark, a 24/7 personal AI agent capable of taking action and integrating with G…

xAI announced the release of Grok 4.3 on its API, claiming it as their fastest and most intelligent model to date, topping leaderboards for agentic t…

Mistral AI announced that its Connectors API is now in Public Preview, allowing developers to register MCP connectors once and reuse them across Le C…

DeepSeek-V4-Pro API is offering a 75% discount until May 5th, 2026, and features an integration update that unlocks 1M context via Claude Code.

vLLM released version v0.21.0, a significant update with 367 commits, which formally deprecates Transformers v4 support and advises users to migrate…