Anthropic to Launch ‘Claude Mythos’ and Upgraded ‘Opus 4.8’ Models in Coming Weeks

Summary: Anthropic is preparing to release the revised ‘Claude Opus 4.8’ model, featuring improved factual accuracy, alongside the powerful ‘Mythos’ model equipped with advanced cybersecurity capabilities, in the upcoming weeks. Under ‘Project Glasswing,’ major technology companies including Amazon, Microsoft, and Apple have been authorized to use Mythos for cybersecurity purposes. Early testers have reported that the new Opus 4.8 model enhances factual reliability by highlighting existing uncertainties and reducing unfounded claims.

Kathmandu, June 15 – Anthropic announced on Thursday that it is set to unveil the updated ‘Claude Opus 4.8’ model soon. Additionally, the company is preparing to launch the robust ‘Mythos’ model for all its customers within a few weeks. Mythos is a large language model (LLM) developed by Anthropic’s AI Lab, incorporating sophisticated cybersecurity features. Reports indicate growing concern among corporate executives and global leaders regarding its potential impact.

As part of ‘Project Glasswing,’ leading technology firms such as Amazon, Microsoft, and Apple have been granted access to use Mythos specifically for cybersecurity applications. According to the company, Opus 4.8 will be available at the same pricing as before, with targeted improvements across various standards aimed at enhancing factual accuracy. Early evaluators have noted that this model is more likely to clearly identify uncertainties in its outputs and reduce unfounded assertions, Anthropic shared.

The creator of Claude stated, “A common issue with AI models is their tendency to rush conclusions without sufficient evidence and to assert progress with full confidence, even when unwarranted.”