Anthropic’s top models return to India after disappointing cutoffs from Fable, Mythos

Anthropic PBC, a company that develops AI tool cloud, announced on Wednesday (July 1, 2026) that it will resume access to Fable and Mythos, two of its most powerful Large Language Model (LLM) models, later in the evening. These models were slammed globally after the United States government issued an export order restricting their use to anyone who was not a US person.

The June 12 ban came just days after Indian organizations (not yet announced) gained access to Mythos, a more advanced model that has not been made available to the public. Anthropic claims that Mythos is adept at cybersecurity tasks, particularly combing large codebases to find vulnerabilities that even human reviewers have not discovered for decades.

Fable is also highly capable of such tasks, but Anthropic has built in protections against multiple use cases. For example, one researcher using Fable reported The Hindu Biology-related work was restricted on the model, with Anthropic saying this was important to prevent the creation of bioweapons.

“Working closely with the government, we trained an improved security classifier that targets and blocks the behavior described in the report,” Anthropic said in a statement, adding that Amazon triggered the US response. “Users will be notified if a request to the Fable 5 is blocked, and the request will instead be sent to Opus 4.8,” the most robust model that was available for most of last month.

sovereignty pain

While the central government is keenly aware of the US-China dynamics centered in frontier AI technology development, the sudden cutback to Mythos has heightened dissatisfaction. During the ban, reports indicated that the government would extend its support to Sarvam AI, a $1.5 billion Bengaluru-based sovereign AI startup with two open source models, by acquiring a small equity stake in the company.

A senior official on Tuesday (June 30, 2026) reiterated what the US government indicated at the Pax Silica summit last week: If access is granted, it will not be taken away. The official said India’s focus on AI – unlike Western economies, which are dealing with employment concerns – will be on deriving maximum utility from AI and delivering widespread “impact”.