Anthropic’s June 2026 launch of the safeguarded Claude Fable 5 for general users and the more capable Mythos 5 for trusted partners established the current Mythos-class baseline, leaving traders focused on whether a follow-on iteration reaches release imminently. With “on or prior to August 25” at 48.5% and “no release by September 30” at 43.5%, the market reflects tight uncertainty around internal training completion, safety classifier deployment, and competitive timing versus OpenAI and Google models. Key swing factors include any official statements on expanded Project Glasswing access, benchmark results demonstrating meaningful gains in cybersecurity or reasoning, and regulatory signals on export controls or misuse risks that could accelerate or defer broader rollout. Historical patterns show Anthropic favors phased, safeguard-first deployments, which traders weigh against pressure to match rival capability jumps.