Claude's Sonnet 5 is designed to accomplish more independently while being more affordable for you.

      Every leading AI lab is competing to demonstrate that their models can operate independently with minimal assistance; pricing is now emerging as the next competitive area.

      Anthropic has recently released Claude Sonnet 5, a model which the company claims performs almost as well as its primary model, Opus 4.8, but at a significantly lower price.

      Introducing Claude Sonnet 5, our most capable Sonnet to date. It can create plans, utilize tools such as browsers and terminals, and operate autonomously at a level that previously required larger and more costly models. pic.twitter.com/UKK8G7ww5h— Claude (@claudeai) June 30, 2026

      So, what’s new about it?

      Sonnet 5 is Anthropic’s most capable Sonnet model to date. It can carry out multi-step tasks, utilize tools like browsers and terminals, and work independently. Achieving this previously required a larger, more expensive model.

      On an agentic coding benchmark, Sonnet 5 achieves a score of 63.2%, a notable improvement from Sonnet 4.6’s 58.1%. Nevertheless, it still falls short of Opus 4.8’s 69.2%.

      In terms of knowledge-work tasks, Sonnet 5 slightly surpasses Opus 4.8, which is somewhat expected, especially since Opus is designed for more challenging decision-making.

      Anthropic

      What are the costs, and is it genuinely safer?

      Sonnet 5 is available today at a rate of $2 per million input tokens and $10 per million output tokens until August 31, 2026. After this date, prices will rise to $3 and $15, respectively.

      This pricing undercuts Opus 4.8, OpenAI’s GPT-5.5, and Google’s Gemini 3.1 Pro, although Gemini 3.5 Flash remains more affordable.

      Regarding safety, Anthropic claims Sonnet 5 is less prone to hallucinations and sycophantic behavior compared to its predecessor. Additionally, the AI model is intentionally weaker in hazardous cybersecurity tasks compared to Opus-class models, which is a deliberate design choice and not an oversight (via Anthropic).

      Nadeem Sarwar / Digital Trends

      Anthropic's Sonnet 5 is now the default model for both Free and Pro plans, available across Max, Team, Enterprise, Claude Code, and the API.

      The launch of Sonnet 5 follows a trend established by competitors: OpenAI’s GPT-5.6 Sol entered preview just last week with subagent task-splitting capabilities, while Google’s Gemini 3.5 Flash, launched in May, is explicitly marketed as agentic rather than conversational.

      Sonnet 5 also features an updated tokenizer that can convert the same input into up to 1.35 times as many tokens as Sonnet 4.6, though the introductory pricing aims to balance this adjustment.