Claude Mythos Preview
Descripción
Claude Mythos Preview is an unreleased general-purpose frontier model from Anthropic, a new tier above Opus (internal codename 'Capybara'). It identified thousands of zero-day vulnerabilities across every major operating system and web browser as part of Project Glasswing, a cross-industry cybersecurity initiative with 12 partners including AWS, Apple, Microsoft, and Google. State-of-the-art on SWE-bench Verified (93.9%), GPQA Diamond (94.6%), USAMO (97.6%), Terminal-Bench 2.0 (82.0%), CyberGym (83.1%), and Cybench (100% pass@1, saturated). Represents a 4.3x increase over the previous trendline for model performance. Deployed under ASL-3 Standard. Best-aligned Claude model to date per Anthropic's risk report, with the first-ever 24-hour internal alignment review before deployment. Not planned for general availability. Pricing for participants: $25/$125 per million tokens (input/output). 244-page system card.
Radar de capacidades
Science usa un proxy de razonamiento cuando los benchmarks científicos dedicados no están disponibles.
Rankings
| Dominio | #Posición | Puntuación | Fuente |
|---|---|---|---|
| Agents & Tools | 3 | 79.0 | LS |
| Multimodal Ranking | 3 | 93.0 | LS |
Puntuaciones de benchmarks (LLM Stats)
Agents
Biology
Code
General
Healthcare
Long Context
Math
Multimodal
Índices de evaluación AA
No hay datos de evaluación AA disponibles
Puntuaciones por categoría LLM Stats
Precios
No hay datos de precios disponibles
Velocidad
No hay datos de velocidad disponibles
Proveedores disponibles
(Unidades internas LS)No hay datos de proveedores disponibles