In a 108K-token JS file memory test, Qwen 3.6 significantly outperformed Gemma 4.
Premium laptop unveiled with Intel Core Ultra and an upgradeable design front and center.
GitHub Copilot has locked individual plans and moved Claude Opus 4.7 up to Pro+.
The targeting and pricing structure of an internal ChatGPT ad pilot has leaked.
A case study of pushing an AST interpreter up to 1.9x faster than CPython through 21 stages of optimization.
The Copilot individual plan is changing, with new sign-ups paused and stricter limits introduced.
Using Hadamard rotation and a Lloyd-Max codebook, the PLI of Gemma 4 E2B was compressed to 2 bits.
An OAuth breach exposed internal access at Vercel and some customer environment variables
SpaceX has signed a deal that would allow it to acquire AI coding startup Cursor for up to $60 billion.
GitHub Copilot's pay-per-use model has ended, so non-subscribers can no longer sign up.
SpaceX has partnered with Cursor and secured an option to acquire it for $60 billion.
Anthropic appeared to move Claude Code from the $20 plan to the $100 plan, but quickly reversed it.
Qwen has open-sourced Qwen3.6-27B, a 27B dense multimodal model.
Anthropic appears to be moving to remove Claude Code access from the Pro plan and shift to API-based billing.
Writing the contract first cut costs by 54%, while team agents were more expensive with no quality gain.
OpenAI has unveiled Images 2.0, which strengthens precise text rendering and multilingual support.
Opus 4.7 keeps the same price, but token counts can increase up to 1.35x.
Signs have emerged that Anthropic is restricting Claude Code access for new Claude Pro subscribers.
The OpenClaw tool for Anthropic's Claude models is supported again, supporting the latest features such as Claude 4.6's adaptive thinking and prompt caching.
A collection of 56 software principles spanning teams, architecture, quality, and design.
Google has released LiteRT-LM, its on-device LLM inference engine.
Background job responses and /schedule routine management have been improved.
Moonshot released FlashKDA, which is up to 2.22x faster than Triton on H20.
Compares API, CLI, and MCP as ways for agents to integrate with external systems, and explains the importance of MCP in cloud environments.