Anthropic released Claude Sonnet 5 on June 30, 2026, and deployed it the same day as the default model for all Claude Free and Pro accounts. The company describes it as the most capable model in the Sonnet family, with performance approaching its flagship Opus 4.8 at substantially lower cost.
Built for agentic work
Sonnet 5 is designed for multi-step, autonomous operation — sustained planning, tool use, and direct control of browsers and terminals. Anthropic says earlier Sonnet models would “stop short” on complex tasks that Sonnet 5 can now complete end-to-end, making decisions across many sequential steps without human course-correction in between.
Compared with Sonnet 4.6, the new model shows measurable improvement in reasoning, coding, and knowledge work. Anthropic also reports lower rates of sycophancy and hallucination, stronger refusal of malicious requests, and better resistance to prompt-injection attacks.
Pricing and developer details
Sonnet 5 launches at an introductory price of $2 per million input tokens and $10 per million output tokens, valid through August 31, 2026. Standard pricing from September 1 is $3 per million input tokens and $15 per million output tokens.
A new tokenizer bundled with the model increases token counts by roughly 1.0 to 1.35 times depending on content type. Anthropic raised rate limits across Claude apps and the API to offset the impact, and set introductory pricing to keep costs roughly neutral for most existing workloads.
The model is accessible via API using the identifier claude-sonnet-5 and is also available on Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry. All Claude plans — Free, Pro, Max, Team, and Enterprise — as well as Claude Code include access.