Claude Opus 5 Is Highly Capable, But Is No Mythos
Zvi's AI Roundups · Zvi Mowshowitz · 2026-07-28
Zvi Mowshowitz evaluates Claude Opus 5 and finds it matches Fable 5 performance on most real-world tasks at half the API price, but falls short of Mythos-class intelligence and draws polarizing reactions for its verbose, confrontational communication style.
Appears in
Extraction
Topics: claude-opus-5llm-benchmarksmodel-evaluationagentic-aiai-model-comparison
Claims
- Claude Opus 5 achieves roughly comparable performance to Fable 5 on most real-world tasks while costing half as much per API token.
- Opus 5 lacks the ability to autonomously chain together seemingly unrelated exploits or reason globally, keeping it below Mythos class.
- Opus 5 excels as a subagent for well-defined bounded tasks but struggles as an orchestrator without human or model supervision.
- Unnecessary refusals are down approximately 85% for Opus 5 compared to Fable 5, making it better suited for scientific research and other classifier-sensitive domains.
- A significant portion of users strongly dislike Opus 5's communication style, describing it as verbose, repetitive, and prone to 'Claude slop' ticks.
Key quotes
It is still not Mythos class. Fable is your only Mythos-class option. Opus 5 does not have The Juice, the ability to autonomously string together a bunch of seemingly unrelated exploits, which extends to other domains, or as much pure intelligence.
Opus 5 seems good at following instructions if and only if it is kept on track, which explains different harnesses causing different takes there.
An unusually large number of people strongly dislike talking to Opus 5. They are sick of the Claude slop, the repetitions of the same ticks and the endless overly complex sentences.