tech
Anthropic debuts Claude Sonnet 5 for everyday agent tasks with lower cyber risk
It says the model can use browsers, plan, code and do knowledge work while posing fewer risks than Mythos and Fable.
TL;DR
- Anthropic releases Claude Sonnet 5, a lower-priced model for everyday agentic AI capabilities.
- Sonnet 5 can handle autonomous tasks like browser use, planning, coding, and knowledge work with fewer cyber risks than Opus and Mythos.
- The model's performance approaches that of Opus 4.8, Anthropic's most advanced widely available model.
- Sonnet 5 was not trained on cybersecurity tasks and has a much lower ability to perform dangerous cyber activities.
- Anthropic is in discussions with the Trump administration regarding model releases, including Sonnet 5.
- More powerful models, Mythos and Fable, are being gradually restored after being taken down for security concerns.
- Companies like Anthropic and OpenAI are focusing on AI agents for task delegation rather than just chat.
- Sonnet 5 is now the default model for Claude Free and Pro users and is available to other tiers at a lower price.
- The release occurs amid a focus on AI usage costs, with some pivoting to cheaper Chinese models.