ModelsAug 23, 2026 · 03:426 sources

Anthropic Says Claude Code 10 Out of 100 Effort Value Is Display Bug Not Stealth Downgrade

From huggingnews.com

Image: huggingnews.com
AI;DR

Anthropic said a numerical mapping error in internal API serving configuration tests caused a display bug showing Claude Code 10 out of 100 effort value, not a stealth downgrade.

Read the sourceSign in to suggest

Sources

+ 5 more
  1. SUPPORT@argofowl“since 2.1.237 the model reads "high" effort as 10 out of 100”x.com
  2. SUPPORT@kimmonismus“They dumbed down fable without telling it seems.”x.com
  3. SUPPORT@trq212“working on rolling out something to make it much easier to report any feedback or regressions”x.com
  4. SUPPORT@kimmonismus“most models currently feel significantly dumber to use. Especially Opus.”x.com
  5. SUPPORT@trq212“Opus 5 is a really spiky model”x.com

One ranked edition, every morning

The stories that matter today. Free, cancel anytime.

Telegram @aidr_today