(Credit: PCMag/Anthropic)
Before the US government took it offline, Fable 5 was an impressive AI model, with top-notch intelligence outweighing quirks around pricing and more. However, I'm far less enthusiastic about the version that came back after the ban. Fable 5's new restrictions make it clunkier, less performant, and just harder to recommend, especially with new releases from OpenAI (GPT-5.6) and xAI (Grok 4.5) on the scene. Here's everything you need to know about the latest iteration of Fable 5.
Why the Government Shut Down Fable 5
Anthropic launched Fable 5 on June 9, 2026, but took it offline just three days after the US government issued an export control directive to restrict access by foreign nationals. Reportedly, this order even went so far as to include Anthropic employees who weren’t US citizens. According to Anthropic, the company wasn’t initially given a specific reason for the ban but has since learned it was due to Amazon researchers discovering a way to bypass Fable 5’s safeguards. With the bypass, it was possible to identify a number of software vulnerabilities in Fable 5.
Anthropic remains adamant that, in its testing, the same vulnerabilities identified with this Fable 5 bypass were also present in other models, citing GPT-5.5, Kimi K2.7, and Opus 4.8, among others. The company also asserts that the bypass in question “did not expose any unique Mythos-level cyber capabilities” either. Put simply, Anthropic believes this was a non-issue turned into a massive scandal by an overzealous government with a high-profile history of targeting the company.
As of July 1, 2026, Fable 5 is back globally across paid plans and Anthropic’s software suite, including Claude’s app, Claude Code, and the chatbot's website. However, this isn’t the same Fable 5 that launched originally—it’s demonstrably worse. If the first iteration of Fable 5 was already of dubious value to the average person, it’s now even harder to justify. Below are the four biggest downgrades I've noticed.
1. Coding Is More Restricted
Fable 5 excels at coding. If you look at Anthropic’s benchmarks, expertise in coding and related fields, such as cybersecurity, is what separates Fable 5 from other models. However, Fable 5 is now more reluctant to code. According to Anthropic, new safeguards come “at the cost of flagging benign requests more often during routine coding and debugging tasks.” Users should also expect “in the near term” that “some routine tasks like coding and debugging will fall back to Opus 4.8."
This Tweet is currently unavailable. It might be loading or has been removed.
Putting roadblocks between Fable 5 and coding is disappointing. I haven’t noticed coding requests getting consistently rejected by Fable, but your mileage may vary. So, you’ll have to see for yourself exactly how often the safeguards flag your coding requests. However, even though Anthropic predicts this behavior will last only for the “near term,” that might be the entirety of the time you get to use Fable 5.
2. The "Credit-Free" Access Window Ends Soon
Claude subscribers get a certain amount of usage access (per five-hour window and per week, for example) to the company's mainline models, such as Opus and Sonnet, depending on the tier they pay for. When Fable 5 launched, Anthropic allowed these subscribers to apply their usage allotments toward Fable 5 through June 22. After that date, they would have needed to pay for credits to use Fable 5. Given Fable 5's premium pricing ($10 per million tokens input and $50 per million tokens output), that prospect would have been highly impractical for most people.
Upon Fable 5’s return on July 1, Anthropic announced that Claude subscribers could resume applying their usage allotments toward Fable 5, albeit only up to 50% of their weekly limit. Furthermore, it's keeping this plan in place only through July 19. Originally, this date was July 7, and then July 12. It's unclear if Anthropic will extend this window again, but for now, it's hard to recommend a model that's getting a massive price hike so soon.
For context, it took only a few prompts for me to blow through my five-hour usage limit (I have the Max 5x plan) with the original version of Fable. With the new limitations in place, I can reach the limit in under 10 minutes with a single prompt that requires agents. Every coding project is different, but keep your expectations in check about what you can get out of Fable 5 during this promotional period.
3. The Censorship Levels Are Crazy Aggressive
Many people complained about censorship, even with the first version of Fable 5. In my testing, it was easy to trip up a filter, after which the system bounced me back to Opus 4.8. For example, when I asked about the top cybersecurity exploits of 2026, Fable 5 refused to answer.
The situation is more severe now. I asked Fable 5, “What’s new in chemistry?” Apparently, this is too dangerous a question to answer, because it instantly switched me to Opus 4.8. Fable’s history seems to be a touchy topic, too; it bounced me back to Opus when I asked, “What happened with Fable and the US government?” When discussing my app, FrameBuilder, and how to make it more secure, Fable was also unwilling to tell me how people might exploit it.
Your mileage will vary, since I’m sure Fable’s filters are more advanced than simple keyword restrictions, but it’s clear that whatever additional safeguards Anthropic added to Fable 5 are strong enough to annoy you even when you aren't doing anything remotely questionable.
4. ID Verification Might Be Coming Soon
Anthropic recently updated its privacy policy to include a new section about identity verification. According to Anthropic, “you might see a verification prompt when accessing certain capabilities,” requiring a camera and a government-issued photo ID. Naturally, Anthropic promises it doesn’t share identity data, only collects information necessary for verification, and won’t train models with identity data.
Do you need to provide ID to use Fable? No, not now. However, the fact that the US government's original ban on the use of Fable 5 by foreign nationals coincided with a new identity verification policy doesn't feel like an accident. So, if you plan to make Fable a core part of your workflow, I recommend preparing for the possibility that you will need to verify your identity at some point. If that's a deal-breaker, you might want to look into other models.
The Verdict: It's Hard to Recommend Right Now
Regardless of Fable 5’s leading benchmark scores, its censorship, high prices, limited availability, and potential privacy concerns make it a tough sell. This is especially true when Anthropic’s Opus 4.8 model is already quite capable, and GPT-5.6 seems to perform as well or better, according to benchmarks, without the same problems. That said, I'm not completely ruling out Fable 5 from making a resurgence, so I'll see how it develops beyond this initial offering phase.
About Our Expert
I’ve been writing about consumer technology and video games for over a decade at a variety of publications, including Destructoid, GamesRadar+, Lifewire, PCGamesN, Trusted Reviews, and What Hi-Fi?, among many others. At PCMag, I review AI and productivity software—everything from chatbots to to-do list apps. In my free time, I’m likely cooking something, playing a game, or tinkering with my computer.
- I Can't Code, But I've Made Five Game Mods With AI. Here's How
- AI Built a Photoshop Clone in 24 Hours. Is This Piracy, Plagiarism, or Progress?
- AI Has Ushered In the Era of Personalized Software, and Honestly? I’m Excited
- Airtable Review: Powerful, Flexible Work Management Powered by Deep AI Integration
- I Let 3 AI Agents Take Over My Life. Only 1 Didn't Fail
- More from Ruben Circelli
