Log inSign up
Log inSign up
Alex
542 posts
@theAlexFerrari

Alex

@theAlexFerrari
infovore/LLMs whisperer/"his mind teeming"
The Hague, The Netherlands
github.com/alexferrari88
Joined May 2020
959 Following
247 Followers
RepliesRepliesRepostsRepostsMediaMedia

Log in or sign up for X

See what’s happening and join the conversation

Continue with phone
or
Log in with username or email
Terms·Privacy·Cookies·Accessibility·US TIDA·Ads Info·© 2026 X Corp.
  • @theAlexFerrari
    Alex
    @theAlexFerrari
    Sep 22
    In my ChatGPT App right now (Windows). It's happening!!
  • @theAlexFerrari
    Alex
    @theAlexFerrari
    Aug 26
    Don't escalate a failed LLM session to a stronger model mid-task. The flawed reasoning trace poisons the smarter model's output. Start fresh with a clean prompt instead.
    arXiv logo
    arxiv.org
    The Handoff Tax: Continuing Non-Native Trajectories in LLM Agents
    Coding agents perform long-running tasks spanning dozens of model calls, tool uses, and code edits. As these runs unfold, users face a practical cost-quality trade-off: escalating to a stronger...
  • @theAlexFerrari
    Alex
    @theAlexFerrari
    Aug 22
    given the high error rate of tool calling from 0x Alpha, my money is on gemini
  • @theAlexFerrari
    Alex
    @theAlexFerrari
    Aug 14
    gpt-5.6-sol re-creating the universe from first principles to fix my bike
  • @theAlexFerrari
    Alex
    @theAlexFerrari
    Aug 9
    - coordinator - architect - frontend/designer - builder - ops - tester - personal operator - researcher - marketer - writer Each fine tuned with SOUL.md, different model+reasoning effort, tool access
    @tonysimons_
    Tony Simons
    @tonysimons_
    Aug 8
    How many Hermes profiles are you running, and what role does each one handle?