Anthropic and OpenAI face rumors of unreleased artificial intelligence models

Unconfirmed rumors and technical leaks suggest that Anthropic and OpenAI are preparing to release new artificial intelligence models, Decrypt and Crypto Briefing reported. Decrypt said users on X observed Anthropic quietly testing unreleased versions of its Fable and Mythos lines, while OpenAI appears to be preparing the launch of GPT-6.2 Astra. Neither company has officially confirmed the developments. [1][4]

According to Decrypt, the speculation is supported by showcase demos, a catalog entry, and a hidden website label that were removed shortly after appearing online. Users reported that Claude was routing certain requests to a stronger unreleased model, and a Mythos 5.5 label was identified in the site code for Anthropic. Decrypt noted that Fable represents the top commercial tier of Claude, which Anthropic sells for writing software and handling office tasks. [1]

Mythos is a restricted model offered to a select group of clients, including government entities, Decrypt reported. Claude Mythos represents the most complicated set of models in the Claude product line. Under Project Glasswing, select companies were granted access to Claude Mythos to scan for security vulnerabilities in critical software because the model was not released publicly. [1][6]

Crypto Briefing and Decrypt both noted that OpenAI is reportedly eyeing the deployment of GPT-6.2 Astra. TestingCatalog AI News reported separately that Anthropic is preparing to release the Claude Mythos 5.5 model. [1][3][4]