Anthropic extends Claude Fable access after GPT-5.6

Anthropic extended Claude Fable 5 access through July 19 after OpenAI’s GPT-5.6 launch.

Anthropic has extended access to Claude Fable 5 for subscribers for a second time in one week, and the timing says a lot about how fast the AI model race is moving. The new extension lasts through July 19 and follows OpenAI’s launch of GPT-5.6 and its flagship model Sol.

The move is simple on the surface: more access, more usage, more attention. Underneath, it is a response to a rival product launch that OpenAI says sets a “new standard” across coding, knowledge work, cybersecurity, and science.

Anthropic is turning access into a product signal

Anthropic announced the extension on X, in an email to subscribers, and in an updated support page. That matters because this is not a quiet backend tweak. It is a public message that Claude users should keep trying Fable 5, compare it with OpenAI’s latest release, and decide which model earns their trust.

The company also said subscribers can still use Fable after they hit the expanded limit if they buy usage credits. That keeps the model in front of paying users even after the promotional bucket runs dry.

Anthropic first widened access on July 7, when it let Claude subscribers spend up to 50% of their weekly subscription limits on Fable. This second extension pushes that window further, and it does so right after OpenAI’s latest launch grabbed the spotlight.

• Anthropic is using access windows as a way to keep Claude users inside the product.
• OpenAI’s launch created a direct comparison point for coding and research tasks.
• The extra credits option means heavy users can keep testing Fable without waiting for the promo to end.

OpenAI’s Sol changed the timing

OpenAI launched GPT-5.6 just three days before the extension, with Sol as the headline model. The company said Sol posted state-of-the-art results in coding, knowledge work, cybersecurity, and science, while also outperforming previous and competing frontier models at a lower cost.

That claim matters because Anthropic and OpenAI are now fighting on the same scoreboard. OpenAI also said Sol and its other new models beat Fable in coding tasks and that Fable failed to test for questions involving scientific research. Anthropic replied that Fable would “fall back” to a previous model for most biology and chemistry requests.

“At the end of the day, why argue with someone when you don’t have the same vision and you don’t trust them?” — Dario Amodei, CEO of Anthropic

The rivalry is personal as well as technical. Anthropic CEO Dario Amodei spent nearly five years at OpenAI before leaving to co-found Anthropic in 2021. He said last month that he left because of distrust in OpenAI CEO Sam Altman. Altman, for his part, criticized Anthropic earlier this year over a Super Bowl ad and called the company “so clearly dishonest.”

Mythos is the bigger story behind Fable

Fable is only the public-facing slice of Anthropic’s more restricted Mythos model. Anthropic says Mythos will become available more broadly in the future, but today access is limited to about 150 organizations across more than 15 countries.

That limited rollout is important because Mythos is the model Anthropic says could be used to find and exploit software vulnerabilities more effectively than any other model, and better than all but the most skilled human security experts. In other words, Fable is the safer wrapper, while Mythos is the more powerful engine behind it.

Anthropic released Mythos last month to a small number of partners for defensive cybersecurity work. The company is careful about where that model goes, and the restricted distribution explains why Fable gets the public attention while Mythos stays mostly behind closed doors.

• Mythos is currently limited to roughly 150 organizations.
• Those organizations are spread across more than 15 countries.
• Anthropic says the model is aimed at defensive cybersecurity use.

What this says about the model wars

There is a pattern here: when one lab ships a high-profile model, the other answers quickly with pricing, access, or messaging. Anthropic’s decision to extend Fable 5 twice in a week is less about generosity and more about keeping Claude in the conversation while OpenAI dominates the news cycle.

For users, the practical question is whether Fable’s extra access changes behavior. If you are already paying for Claude, the answer is yes, because you can spend more time comparing it with OpenAI’s latest releases without burning through your normal weekly cap. If you are a team evaluating models for coding or research, the current split between public access and restricted access makes the decision more complicated.

For now, the clearest takeaway is that Anthropic is willing to use temporary access boosts as a competitive tool. The next question is whether it follows with a larger Fable rollout, or whether the company keeps reserving the most powerful version of Mythos for a narrow set of partners while using Fable to hold the public line.

That answer will matter more than the promotional end date. If OpenAI keeps shipping faster than Anthropic, expect more short-term extensions, more credit offers, and more public one-upmanship around benchmark claims.