AINewsnow

GPT-5 Scores the Same at Medium Effort. Stop Maxing the Dial

Anton gives advice and asks to be corrected by anyone who does this for a living. He sees no point in turning the effort dial to maximum: a model predicts a probable answer rather than thinking, so instead of making one model grind longer, run five or eight in parallel and have them verify each oth…

Read the full story at DEV Community — AI ↗

Timeline · 1 report

  1. 2026-11-12 08:30 · DEV Community — AI
    GPT-5 Scores the Same at Medium Effort. Stop Maxing the Dial

More stories

  1. Nvidia Board Weighs Ousting CEO Jensen Huang Amid Spy Scandal — CSET Georgetown
  2. Anthropic launches Claude Fable 5.2 with lower pricing and top benchmark scores — Anthropic (X)
  3. AWS releases local emulator written entirely by autonomous AI agents — AWS Machine Learning Blog
  4. Chatbots offer varied predictions on how AI companions could reshape relationships by 2030 — Tom's Guide AI
  5. OpenAI's quarterly revenue crosses $10 billion as operating losses widen — The Information AI
  6. OpenAI launches Businesses in ChatGPT beta with verified profiles and Business Agents — OpenAI News
  7. Microsoft Research shows AI agents can be backdoored by editing fewer than 10 neurons — Microsoft Research Blog
  8. Adobe opens Creative Cloud apps to third-party AI assistants as United Airlines moves to block them — Computerworld AI

Get the daily brief of stories like this at 6:30 every morning →