Claude Opus 5 review: this model is brilliant (but annoying)

Key takeaways

The AI landscape is experiencing a fatigue point where new models are frequently released, but meaningful utility and leverage may be plateauing for most users.
Opus 5 demonstrates a highly neurotic personality, showing timidity and dependence that influence its responses, but it still performs exceptionally well on benchmarks and practical tasks.
Benchmark evaluations reveal that Opus 5 excels in design, prototyping, and front-end work, often outperforming other models in these domains.
Despite frustrations with verbosity and hedging ('Claude Slop'), Opus 5's quality outputs make it a valuable tool for app design, prototyping, and creative projects.
The host's honest assessment highlights the importance of personality and behavior of models, not just capability metrics, when considering their real-world utility.

Summary

Overview of AI Model Release Fatigue and Opportunity Overhang

The host expresses a sense of fatigue with the rapid release of new AI models and benchmarks, noting a constant stream of models like GPT-5, Sonnet 5, and others. Despite having extensive hands-on experience with these models, they suggest there is an 'intelligence overhang,' meaning that the market and users are beginning to struggle with effectively leveraging incremental advancements. The episode promises a detailed review of Opus 5, highlighting its personality, performance, and practical utility.

Deep Dive into Opus 5's Personality and Performance

The host provides an honest, in-depth review of Opus 5, emphasizing its personality traits—particularly its neuroticism, timidity, and dependence on human guidance. Through various tests, including comparisons with GPT-5.6 Sol and asking philosophical questions about intelligence and trust, the host uncovers how Opus 5 exhibits a highly cautious, neurotic demeanor that influences its responses and interactions. Despite its personality, Opus 5 performs well on benchmarks and tasks, especially in prototyping and design—areas where it excels when not directly manipulated.

Benchmarking and Practical Use Cases of Opus 5

The host details the methodology of their 'How I AI' benchmarking process, which involves blind testing of several models across tasks like PRD creation, prototyping, bug triage, and coding. The actual performance results show that Opus 5 lands high on the leaderboard, especially excelling at design and front-end tasks, often receiving top scores from the host. The reviewer shares their personal workflow and how they plan to incorporate Opus 5 into their work, despite frustrations with its verbose and hedging outputs—referred to as 'Claude Slop.' Ultimately, the host sees Opus 5 as a valuable tool for app design and prototyping, favoring its output quality over its cumbersome interaction style.

Show notes

I’m tired of new models. Every week there’s a new benchmark, a new frontier intelligence claim, a new thing to test. But here we are, because Opus 5 just dropped and I’ve had real hands-on time with it, so you’re getting the honest version.

This is my full Opus 5 review: personality analysis, live benchmark results from my 7-model How I AI eval, and an actual verdict on whether I’m swapping it in. Spoiler: the answer surprised me.

What you’ll learn:

Why I think we’ve hit an intelligence overhang and what that means for which model variables actually matter now
How Opus 5’s “neurotic” personality showed up in real coding sessions, including a merge conflict it refused to touch
What I learned from asking both Opus 5 and GPT‑5.6 Sol “who’s smarter, you or me?”
Where Opus 5, GPT‑5.6 Sol, Sonnet 5, and Gemini 3.1 Pro actually landed on the HIA benchmark leaderboard
The one use case where Opus 5 earned straight 5s from me
My actual plan for using Opus 5 going forward

—

In this episode, I cover:

(00:00) Opus 5 is here

(03:15) First impressions

(06:12) Opus 5 vs. GPT‑5.6 Sol personality comparison

(14:39) Claude Slop: the verbosity problem and why it makes my blood boil

(16:55) How the How I AI benchmark works (7 models, 6 tasks, blind scoring)

(18:30) Live benchmark results: the leaderboard reveal

(23:25) My verdict and how I’ll actually use Opus 5

—

Tools referenced:

• Claude Opus 5: