tech
Introducing Claude Sonnet 5
Claude Sonnet 5 is built to be the most agentic Sonnet model yet. It can make plans, use tools like browsers and terminals, and run autonomously at a level that, just a few months ago, required larger and more expensive models.
TL;DR
- Claude Sonnet 5 is the most agentic Sonnet model, capable of planning, tool use, and autonomous operation.
- It narrows the performance gap with Opus-class models, approaching Opus 4.8's capabilities at a lower cost.
- Sonnet 5 shows significant improvements over Sonnet 4.6 in agentic performance, including reasoning, tool use, and coding.
- Safety assessments indicate Sonnet 5 has a lower rate of undesirable behaviors than Sonnet 4.6, though it's less capable in cybersecurity tasks than Opus models.
- The model is available across all Claude plans and via the Claude API with introductory and standard pricing tiers.
- Early access partners reported Sonnet 5's ability to complete complex tasks, self-check output, and operate efficiently.
- Safety evaluations show Sonnet 5 is better at refusing malicious requests and resisting prompt injection attacks compared to Sonnet 4.6.
- While Sonnet 5 shows some improvement in developing software exploits, its cyber capabilities are substantially poorer than Opus 4.8 and Mythos 5.
- Cyber safeguards are enabled by default for Sonnet 5 due to its enhanced safety features.