Opus 5 doesn’t follow instructions , it improvises incompetence. Point out its mistake, blames you for the failure, then rage-quits the conversation. as if hallucinating wasn’t enough, it also needed an ego.

There’s simply no reason to pay $20 for Claude Pro when ChatGPT Plus exists. OpenAI has decisively beaten Anthropic & Anthropic clearly doesn’t care about its $20 subscribers anymore. I’d rather pay for two ChatGPT accounts than keep forcing myself to use Claude Pro with Opus 5.

What exactly is the point of a Pro tier if subscribers are left with the substandard and frustratingly incompetent Opus 5, while Fable is locked behind additional usage credits? A pro subscription shouldn’t feel like second-class access.

Pro users are now stuck with Opus 5, while those with the means to afford more get Fable 5.1. That tells you more about Anthropic’s priorities than any mission statement ever could: the wealthy feast in the ivory tower; everyone below gets the scraps.

Anthropic is doing its part to save SWE jobs from AI ,they released Opus 5 that writes enough code to create the illusion of automation, but enough dumb mistakes to guarantee you’ll always need an engineer to understand and fix it.

they made Opus 5 dumb on purpose. Every model distilled from it inherits the stupidity and therefore can’t beat Fable on the benchmark.

Opus 5 is genuinely impressive it takes longer to misunderstand the prompt, burns more tokens explaining why, then confidently delivers exactly what you told it not to do.

@AnthropicAI has not enough compute to give fable and a old style powerful opus together you can only have one. Hence the opus which functions like a toddler

@AnthropicAI kept crying about the capacity and compute required for the demand for fable then they added a fuck all model which can’t do what it’s told called opus 5 and it is so bad people don’t wanna use it hence it allowed more compute for fable. Long game