OPEN LINK ↗
// 2h agoNEWS
Anthropic clarifies Claude Mythos pricing
Anthropic corrected a pricing misconception regarding its unreleased Claude Mythos model, clarifying that the cost is $25 per million input tokens and $125 per million output tokens rather than a flat monthly fee of "$25/m". Claude Mythos is a highly capable frontier AI model with advanced coding and cybersecurity capabilities, currently restricted to vetted defensive partners via the Project Glasswing initiative.
// ANALYSIS
While resolving the ambiguity between "million" and "month" in token pricing is a minor communication fix, the underlying pricing structure reveals the massive premium placed on specialized cybersecurity models.
- –Claude Mythos is priced at 5x the rate of Anthropic's flagship Claude Opus ($5/$25 per million tokens), reflecting the high compute demands and specialized nature of zero-day vulnerability detection.
- –Restricting model access to Project Glasswing partners shows Anthropic's strict dual-use risk management, keeping powerful cybersecurity exploitation capabilities out of the public domain.
- –The use of $100 million in credits suggests that despite the high cost, Anthropic is heavily subsidizing defensive testing to accelerate software vulnerability patching.
// TAGS
anthropicclaude-mythosproject-glasswingai-pricingsecurity
DISCOVERED
2h ago
2026-06-08
PUBLISHED
2h ago
2026-06-08
RELEVANCE
5/ 10
AUTHOR
ollobrains