Anthropic has launched Claude 3.5 Sonnet, the latest addition to its AI model lineup, positioning it ahead of previous models and competitors such as OpenAI’s GPT-4 Omni. Available for free on Claude.ai and the Claude iOS app, the model can also be accessed via the Anthropic API, Amazon Bedrock, and Google Cloud’s Vertex AI. Claude 3.5 Sonnet is priced at $3 per million input tokens and $15 per million output tokens, featuring a 200,000-token context window.

Claude 3.5 Sonnet has set new benchmarks in various domains including graduate-level reasoning (GPQA), undergraduate-level knowledge (MMLU), and coding proficiency (HumanEval). It shows notable advancements in understanding nuances, humor, and complex instructions, excelling in generating high-quality content with a natural tone. The model operates at double the speed of Claude 3 Opus, making it suitable for tasks like context-sensitive customer support and multi-step workflows.

According to Anthropic's internal evaluation, Claude 3.5 Sonnet solved 64% of coding problems in an agentic coding assessment, surpassing Claude 3 Opus, which solved 38%.

The model's capabilities extend to writing, editing, and executing code independently, facilitating tasks such as updating legacy applications and migrating codebases. It performs well in visual reasoning tasks, interpreting charts and graphs accurately, and can transcribe text from imperfect images, benefiting industries like retail, logistics, and financial services.

Anthropic has introduced Artifacts on Claude.ai, allowing users to generate and edit content like code snippets, text documents, or website designs in real time. This feature transforms Claude into a collaborative work environment, with future plans to support team collaboration and centralized knowledge management.

Emphasizing safety and privacy, Anthropic highlights that Claude 3.5 Sonnet has undergone rigorous testing to prevent misuse. External evaluations by experts, including the UK’s Artificial Intelligence Safety Institute (UK AISI), have contributed to refining the model's safety protocols. Anthropic affirms it does not train generative models on user data without explicit consent, incorporating feedback from child safety experts to enhance its classifiers.

Looking forward, Anthropic intends to release Claude 3.5 Haiku and Claude 3.5 Opus later this year, introducing features like Memory to enable Claude to remember user preferences and interaction history.