SynopsisAnthropic has launched Claude Fable 5 and Claude Mythos 5, introducing a new class of AI models designed for long-horizon, autonomous tasks. The release highlights the industry’s shift toward agentic AI, combining breakthroughs in software engineering and scientific research with stringent safeguards aimed at mitigating emerging cybersecurity and biosecurity risks.ET OnlineSan Francisco-based AI company Anthropic has introduced Claude Fable and Claude Mythos 5, a new generation of “Mythos- class” models that signal broader industry shift from conversational AI to autonomous systems capable of executing complex, long-duration tasks. The announcement comes as competition among AI developers increasingly focuses on agentic capabilities rather than conventional chatbot performance. While Claude Fable 5 is being made broadly available through the company’s API, Claude Mythos 5 will be offered under restricted access programmes for selected organizations and government collaborations, including Project Glasswing. According to Anthropic, the new models represent its most advanced AI systems to date. What sets them apart is their ability to sustain reasoning and execute multi-step workflows over extended periods, allowing them to tackle tasks that previously required significant human oversight. Beyond coding, the models have demonstrated notable gains in vision capabilities. Anthropic said the systems can interact with complex video games using only screenshots, without relying on additional game-state information or navigation tools. In scientific research, the company reported that the models have generated novel hypotheses in molecular biology and genomics, with some findings already receiving independent validation. The launch also underscored why Anthropic has adopted a differentiated deployment strategy. The company believes Mythos-class models possess capabilities that could potentially be misused in areas such as cybersecurity and biological research. To mitigate these risks Claude Fable 5 incorporated conservative safeguards that automatically redirect high-risks queries to the older Claude Opus 4.8 model. Anthropic acknowledged that this approach could result in occasional false positives but said the safeguard would be refined over time. As part of its safety framework, the company has also introduced a 30-day data retention policy for Mythos class traffic to study emerging attack patterns and strengthen its protective systems. Both models are priced at $10 per million input tokens and $50 per million output tokens, with access to Fable 5 being rolled out in phases across Pro, Team and Enterprise plans through June 2026.The launch highlights a broader transition in artificial intelligence from systems designed primarily to answer questions to AI agents capable of planning, executing and validating complex tasks with increasing levels of autonomy, while raising fresh questions about how such capabilities should be governed.Nominate for most innovative AI product 2026 Disclaimer Statement: This content is authored by a 3rd party. The views expressed here are that of the respective authors/ entities and do not represent the views of Economic Times (ET). ET does not guarantee, vouch for or endorse any of its contents nor is responsible for them in any manner whatsoever. Please take all steps necessary to ascertain that any information and content provided is correct, updated, and verified. ET hereby disclaims any and all warranties, express or implied, relating to the report and any content therein.Read More News onRead More News on