Anthropic has recently unveiled its latest iteration of AI models, Fable 5.1, which promises to be up to 45% cheaper for complex tasks. The company claims this is achieved through reduced pricing on cached data, making it more cost-effective for businesses. Additionally, Mythos 5.1 is set to launch in Project Glasswing, expanding the company's reach.

Early adopters like Every CEO Dan Shipper and Box CEO Aaron Levie have praised the new models, with Shipper noting that Fable 5.1 is not only more efficient but also more human-like in its communication. Levie added that it has improved its ability to interpret complex data, aligning with his company’s needs.

However, some features, such as the “complete privacy” offered by Enterprise Frontier Safeguards, remain a point of interest. These safeguards store data on customer cloud servers, ensuring data privacy. While Fable 5.1 is now available across all platforms, Mythos 5.1 is only available to select participants in Project Glasswing.

Anthropic also addressed concerns about data retention and overzealous safeguards, stating that Fable 5.1 has more precise safeguards that are less likely to block basic biology questions. Yet, Mythos 5.1 maintains the same biological restrictions as its predecessor.

Despite these advancements, Anthropic still redirects certain cybersecurity tasks to Opus models, such as penetration testing and exploit generation. This move suggests a strategic focus on differentiating between general AI and specialized tasks.