AI models

New model releases and what they change in practice for developers: benchmarks, context windows, pricing, and real-world behavior.

• 5:06

  Stop Letting Claude Code Grade Its Own Work

  2026-09-29

• 7:52

  Jev Won't Make Claude Code Cheaper. Its Own Bench Says So

  2026-09-26

• 11:16

  Spotify Cut Claude Code Tokens 90%. I Rebuilt It With 1 Hook

  2026-09-11

• 11:34

  Fable 5.1 Is The Best Model. The Fine Print Costs You More

  2026-09-04

• 12:15

  Opus 5 Has A Taming Guide Almost Nobody Opened. We Tested It

  2026-08-28

• 10:42

  DeepSeek Built A Claude Code Rival Made Entirely Of Plugins

  2026-08-14