All Posts

In METR's randomized controlled trial, experienced developers using AI tools were actually 19% slower. They believed they were 20% faster.

Anthropic partnered with CodePath to give Claude and Claude Code to 20,000 students for free. It looks like charity. It's a 10-year strategy.

DeepSeek V4 is a 1 trillion parameter model that runs on two RTX 4090s. MoE architecture and MLA compression made it possible. GPT-4 class inference without $50,000 worth of H100 infrastructure.

Anthropic raised $30 billion in its Series G at a $380 billion post-money valuation. Claude Code annual revenue run rate surpassed $2.5 billion, more than doubling since early 2026.

OpenAI announced GPT-5.3-Codex and revealed that "this model contributed to building itself." Recursive self-improvement, a concept that existed only in theory for decades, just became real.