BiRefNet fine-tuned on 1228 anime images becomes ToonOut — how a generic matting model that breaks on anime hair gets repaired, plus BiRefNet's own dynamic-resolution training strategy
Notes from two Fly.io videos — Firecracker microVMs, why fctl is a fat client, and how the Machines platform shares the same primitive between fly launch and direct API users
Anthropic just leased the entire SpaceX Colossus 1 data center and turned it into higher rate limits for Claude Code and the Claude API. 220K GPUs and 300MW translate into doubled Claude Code windows, lifted peak throttles, and big Opus API bumps — but the real news is renting a frontier supercomputer from a direct rival
An analysis of the Zig to Rust porting guide hiding in Bun's claude/phase-a-port branch and the LLM-driven migration workflow it encodes
How Polaris Office's metric-compatible font generator dodges font licensing pain and what an adjacent LLM-as-Judge rubric thread reveals about production AI evaluation
How the public-apis curated free API list is being rediscovered as the inventory layer for the MCP catalog era
A walk through Microsoft's six-paper LLMLingua line from EMNLP 2023 through CoLM 2025 covering prompt compression KV-Cache optimization long-context acceleration and why production adoption lags the research
A digest of three arxiv papers surfacing within days of each other read along three axes — cooperation persistence and structure
How OpenAI runs Realtime voice traffic on Kubernetes by separating a stateless global UDP relay from stateful WebRTC transceivers
OpenAI Codex moves inside ChatGPT plans, unifying CLI, app, IDE, web, and a new Python SDK under one ChatGPT account — with policy clauses and workflow recommendations
