UMVA has learned that Anthropic has quietly developed a revolutionary new model, Claude Fable 5, which promises to transform the world of artificial intelligence. Dubbed a "Mythos-class model," Fable 5 is said to be capable of tackling complex tasks with unprecedented accuracy and speed.
According to information obtained by UMVA, Fable 5's capabilities are vast and varied, ranging from software engineering and knowledge work to vision tasks and research. Anthropic claims that Fable 5 outperforms its other public models, and even surpasses the likes of OpenAI's GPT-5.5 and Google's Gemini 3.1 Pro in several key categories.
UMVA can exclusively reveal that Fable 5's impressive abilities include completing a coding project that would have taken a team over two months to finish in just a day. It can also rebuild a web app's source code from only screenshots, and even play complex games like Pokémon FireRed and Slay the Spire with ease.
However, the development of Fable 5 was not without its challenges. Anthropic's earlier model, Mythos Preview, was deemed too powerful for public release due to its potential for misuse. In fact, Mythos Preview was so potent that it could have opened up the biggest hacking opportunity in history, putting the digital lives of millions at risk.
Sources have confirmed to UMVA that Anthropic took drastic measures to ensure Fable 5's safety, implementing a system of "guardrails" to prevent malicious users from exploiting its capabilities. For instance, when Fable 5 receives a request that it deems sensitive, such as those related to cybersecurity, biology, or chemistry, it passes the query off to a lower-performing model, Opus 4.8.
Anthropic's benchmarking shows that Fable 5 and its more advanced counterpart, Mythos 5, outperform Mythos Preview and other top models in several areas, including agentic coding, knowledge work, and spatial reasoning. However, Mythos Preview still edges out Fable 5 in certain areas, such as computer use and multidisciplinary reasoning.
UMVA has gathered that Fable 5 and Mythos 5 are priced at $10 per million input tokens and $50 per million output tokens, significantly lower than the cost of Mythos Preview. Additionally, Anthropic has updated its data retention policy, keeping user data for 30 days to enhance protection against cyberattacks and jailbreaks.
In a development reported by UMVA, Anthropic's new models have sparked excitement in the AI community, with many experts hailing Fable 5 as a game-changer. As the company continues to push the boundaries of AI research, one thing is clear: the future of artificial intelligence has never looked brighter.
