Anthropic Fixed Fable 5.1's Worst Problems...So Why Am I Staying With Opus?
Fable is perhaps the most intriguing AI model to date. Anthropic positioned it as the world's most powerful LLM, comparable only with its big sibling, Mythos, which was too powerful for a public release . Then the US government banned it almost immediately, only to allow it to return in a less capable state . Now, Fable 5.1 is here. I've tried it, and I find it far more compelling, thanks to its friendlier pricing, relaxed censorship, and more. Here are the top five things Fable 5.1 gets right—and why all of these massive upgrades still aren't quite enough to change my daily routine.
1. Fable 5.1 Is Cheaper, Unlocked, and Here to Stay
Previously, a cutoff date when Fable would only be accessible with purchasable credits rather than a regular Claude subscription loomed large over the experience, but that's no longer the case. Now, you can access Fable via Enterprise, Max, and Team plans for up to 50% of your weekly usage . Of course, you still have the option to buy credits to keep using Fable beyond those limits.
Fable 5.1 is also cheaper if you buy credits. According to Anthropic , you can expect savings of up to 25% in general and up to 45% for highly agentic work. This comes courtesy of reduced prices on cache reads, "where the model reads inputs that have already been processed and stored." If you want to use Fable 5 heavily, these savings could potentially become significant.
Fable might not cost quite as much as before, but it's still Anthropic's most expensive model
Just keep in mind that Opus 5 is still significantly cheaper per token and is available on all paid Claude plans.
2. Generous Usage Limits (With a Fine-Print Catch)
Anthropic is currently (and has been since May) running a promotion that lets you enjoy 50% more weekly usage with Claude Code. However, when that promotion expires, Anthropic promises a permanent 25% increase to weekly usage across Enterprise, Max, Pro, and Team plans. So, although this means you can expect a 25% usage decrease versus what you're getting now, that's a permanent 25% bump over the baseline. Every bit of Fable usage counts, given the weekly usage limits.
3. Goodbye, Overly Aggressive Censorship
Fable 5's censorship was incredibly aggressive. Previously, when I asked benign questions such as, "What happened with Fable and the US government?" or "What's new in chemistry?", Fable bounced me back to Opus. And I couldn't ask about potential exploits to any coding projects I was working on, either. Now, Fable 5.1 handles all of these prompts without issue. Safeguards are still in place, though, so don't expect to hack into any mainframes with it.
Fable 5.1's censorship filters don't pop up nearly as often with basic prompts
When Fable 5 came back online after its ban, Anthropic noted that it might falsely flag coding tasks. That tendency limited the LLM as a coding tool, but the company promised to continue refining its classifiers behind the scenes. It seems to have followed through on this initiative, as I haven't encountered any issues with Fable 5.1 blocking coding-related prompts. Of course, your mileage may vary, but I no longer have any meaningful concerns about censorship.
4. Surprisingly, You Don't Need an ID Check
When Fable 5 returned after its prohibition, Anthropic also updated its privacy policy to include a new section on ID verification . Considering the US government's Fable ban focused on its use by foreign nationals, it seemed possible that you would soon need to provide ID to use Fable. However, this hasn't been the case. Although Anthropic notes that "a few use cases…might see a verification prompt when accessing certain capabilities," this isn't standard policy, and you can safely use Fable without the need to upload a picture ID.
5. Fable Reclaims Its Benchmark Crown From Opus
Fable 5 was only briefly Anthropic's most capable model. Opus 5 quickly outpaced it not just in efficiency but also in performance across Anthropic's benchmarks for a variety of use cases, from agentic coding to biology. Since you could use Opus 5 with any plan to the full extent of your weekly usage limit, Fable 5 became difficult to justify.
Fable 5.1 scores well on Anthropic's benchmarks, putting it ahead of Opus 5
Now, Fable 5.1 tops Anthropic's list . So, if you actually need best-in-class performance, you should consider using Fable 5.1 first. That applies to everything from agentic coding tasks to complex business workflow development to scientific research and more.
Why I'm Still Sticking With Opus
Fable 5.1 is objectively much better than the original. If you have a tough task that Opus 5 just can't handle, you can confidently turn to Fable 5.1 without spending extra money (or as much if you opt for credits), verifying your ID, or worrying about censorship. However, chances are that you won't notice much of a difference versus Fable 5 or Opus 5 for simple coding projects or web searches.
As always, I plan to keep testing Fable to see if any aspect of it could convince me to switch from Opus, but I haven't found one yet.
PCMag and Yahoo may earn commission from links in this article.
