Anthropic released Claude Opus 5.5 on September 22, the same day OpenAI released GPT-6 Sol and Luna. OpenAI is making its lower-end models cheaper, while Anthropic is still pushing upward. On price to performance, Opus 5.5 now beats Fable 5.1, which makes one wonder who would still use Fable.

On Artificial Analysis’s Intelligence Index, Fable 5.1 tops out at 53, costing $5.98 per benchmark task at extra-high effort. GPT-6 Astra already matched that 53 for $3.26. Opus 5.5 does both: at high effort, it scores 54 for $1.82, more heft for under a third of Fable’s price. Even Anthropic’s announcement says Opus 5.5 “performs at the level of Claude Fable 5.1 on most work”. By “most work”, I believe the company means Fable now excels only in some specialized areas.

OpenAI’s releases work the other end. Opus 5.5 at low effort scores 42 for $0.55, while GPT-6 Sol at high gets a comparable 43 for $0.37. GPT-6 Luna at max scores 37, just under Sonnet 5’s 38 at max, for $0.07 a task against Sonnet’s $5.09. There is no GPT-6 Terra, and I wonder if Terra is being phased out entirely. Luna, Sol, and Astra sit about 10 points apart, especially at medium effort, where they score 29, 40, and 50. That leaves Terra in an awkward place, maybe as an all-arounder if it ever does land.

On Anthropic’s side, Haiku never got a 5.0 and still sits at 4.5, scoring 17 for $0.21 a task. At that price, it’s cheaper to just use Luna everywhere. Anthropic says Sonnet 5.5 and Haiku 5.5 will follow in the coming weeks, but I don’t know where Haiku’s price will land. With OpenAI pushing the lower end this competitively, I’m not sure where that leaves Anthropic.

I’ve used Opus 5.5, and so far it’s simply the best Claude offers, with nothing else in the lineup reaching its scores for less. Not to mention Opus 5.5 isn’t tied down by a 50% meter the way Fable is. On the Codex side, I am still running some tests before I can move my jobs over to GPT-6 models. My last model test pinned down the Index score one of those jobs needed, which makes it easy to find the configuration that balances a better model and a better price.

Different people have different jobs, and different jobs need different Index scores. I’m curious if Opus 5.5 simply phases out Fable. On price to performance alone, I can’t see who would still pick Fable 5.1.