Claude Opus 5 Brings Fable-Level Intelligence at Half the Price
Published on
Claude Opus 5 literally JUST came out of the oven, and apparently it’s even better than Fable/Mythos.
This is Anthropic’s fourth launch in less than two months: Mythos Preview, Sonnet 5, Fable 5 and now Opus 5. If you think it’s hard to keep up with this pace, imagine what it’s like for us at IA Sob Controle 😅
The race between the AI labs left the “one bombshell launch every six months” mode a while ago and moved to a cycle of fast improvements in capability, cost and speed. And that changes how those of us who use these tools day to day need to think about which model to pick for each task.
In the benchmarks below, Opus 5 shows up as the new state of the art in some programming and general knowledge tests. On Frontier-Bench v0.1, focused on software engineering, Opus 5 more than doubled Opus 4.8’s performance, came in 10% ahead of Fable 5, and still came out cheaper per task.
On ARC-AGI 3, one of the hardest benchmarks out there, focused on super complex puzzles for AIs, it scored three times what GPT-5.6 Sol did. And that’s a benchmark that had been kind of stuck for a while.
However, according to Anthropic itself, it still falls behind Fable/Mythos 5 on cybersecurity tasks.
They also said the model’s safety classifiers are triggered 85% less often than on Fable 5. Great news, because Fable’s are super annoying, full of false positives. In other words, Opus 5 gets blocked less on legitimate tasks and still beats Fable on a good chunk of the programming tasks that don’t involve cybersecurity.
As for the rest: the pitch is that it gets really close to Fable 5’s intelligence for HALF THE PRICE. While Fable 5 charges US$ 50 per 1M output tokens, Opus 5 charges US$ 25.
China’s Moonshot AI had been bothering the American labs when Kimi K3 got right up next to Fable’s and GPT Sol’s numbers, offering similar intelligence for a much lower price: US$ 15 per 1M output tokens. Opus 5 pushes Fable’s high price down again, and shows more and more that the race stopped being just “who’s smarter” and became “who delivers enough intelligence at the lowest cost per task”, which is a much tougher fight to win.
Worth noting that Fable 5 is still the one Anthropic recommends, at least for planning longer, more complex autonomous tasks. Opus 5 is the everyday worker, which handles most of the real programming and knowledge work without demanding top-of-the-line prices.

