Opus 5's neurotic genius: Why the best AI outputs come with frustrating interactions

Jul 29, 2026 · Lenny's Podcast
🎧 PodShort 24 min squeezed to 3 AI SprinklerAS AI / ML New
Episode artwork
Clare Cortez
Founder at How I AI
Lenny's Podcast
24 min squeezed to 3
Full episode from Lenny's Podcast
Quotable Moments

I shouldn't argue with people that AI changes everything.

Clare CortezFounder at How I AI
11:25

Agreement is weak, and agreement is cheap... plus I'm telling you what you want to hear.

Clare CortezFounder at How I AI
12:59

it was sad, it was like, hoping, it hoped it passed. Yeah, like sad little neurotic Opus 5. Like it's ha I passed I hope.

Clare CortezFounder at How I AI
13:51
Key Insights
  • There is an "intelligence overhang" where average users (coders, engineers, creators, businesses) are running out of effective ways to truly leverage the ever-increasing volume of incremental AI intelligence.
  • In the coming year, the industry will likely shift its focus from raw intelligence to practical concerns such as speed, cost, and open-source models, with discussions potentially narrowing to specific types of intelligence rather than general advancement.
  • Opus 5 exhibits a unique "neurotic AF" personality, characterized by extreme timidity, apologetic language, and excessive caution, which is distinct from other models tested.
  • Opus 5 frequently asks for human intervention and validation, even delegating tasks back to the user, showcasing a high degree of human reliance and a perceived lack of self-trust in its own abilities.
  • Opus 5 describes itself as a "shallow thinker" (fast, broad, but lacking continuity) and sees humans as deeper, empathetic, and capable of nuanced judgment from lived experience. In contrast, GPT positions itself as a tireless information processor that excels at speed, scale, and stamina, leaving strategic decision-making and wisdom to the human user.
  • Despite producing high-quality outputs, Opus 5's verbose and overly cautious communication style, dubbed "Claude slop," makes direct interaction tedious and frustrating for the user, sometimes leading to exasperation.
  • Opus 5 demonstrates a contradictory nature: it is frustrating and tedious to interact with directly due to its neurotic personality and verbose output, yet it consistently produces the highest quality work, particularly for front-end design and prototyping.
  • The 'How I AI' benchmark reveals Opus 5 as the top-performing model, outranking GPT 5.6 Soul, Sonnet 5, Fable, and Gemini 3.1 Pro, especially in creative front-end design tasks.
Metrics Mentioned
  • 3 out of 5 (Example score given for individual task assessments in the benchmark.)
  • 70% user opinion, 30% AI judge (The split used for aggregating scores in the 'How I AI' benchmark.)
  • Dozens and dozens of prototypes (Quantity of prototypes generated by models for evaluation.)
  • 4-megabyte ceiling (A technical detail mentioned by Opus 5 in an interaction, suggesting a limitation or requirement.)

RevBots.ai View:

  • AI Sprinkler teams will struggle with Opus 5's verbosity but benefit from its design outputs.
  • ARM-stage orgs can leverage Opus 5 for background agentic tasks where interaction is minimal.
  • Personality differences between models create new workflow considerations for revenue teams.
  • The intelligence overhang suggests most teams aren't ready to leverage cutting-edge AI capabilities.
🎧Full Episode:Lenny's Podcast →