Anthropic unveils Claude Opus 4.8, its most honest model and smartest coder yet

Anthropic released Claude Opus 4.8, its flagship model, with better coding, reasoning, and agentic task performance.

The update also improves self-checking and reduces deceptive or unsupported answers, aiming for greater honesty and reliability.

Anthropic added Dynamic Workflows for Claude Code, an effort-control setting, and lower fast-mode pricing.

More advanced cybersecurity-focused models are still in testing, including Project Glasswing and Mythos-class work.