Anthropic Launches Claude Sonnet 5 to Cut Costs on Autonomous AI Agents
Anthropic has released Claude Sonnet 5, a midsize model designed to execute complex, multi-step agentic workflows at a fraction of the cost of flagship models.
- Claude Sonnet 5 is now the default model for Anthropic’s free and Pro plans, offering autonomous planning, browser, and terminal tool use.
- Introductory API pricing is set at $2 per million input tokens and $10 per million output tokens through August 31, 2026, before rising to $3 and $15 respectively.
- On agentic coding benchmarks, Sonnet 5 scored 63.2%, outperforming its predecessor Sonnet 4.6 (58.1%) and approaching the flagship Opus 4.8 (69.2%).
Anthropic has released Claude Sonnet 5, a beefed-up iteration of its midsize model engineered specifically to run autonomous AI agents more cheaply. Available immediately as the default model for both free and Pro subscription tiers, the model is designed to plan, use browsers or terminals, and execute multi-step tasks without constant human intervention.
According to Anthropic, the launch aims to lower the financial barrier for developers building agentic workflows. For early adopters, Anthropic is offering introductory API pricing through August 31, 2026, at $2 per million input tokens and $10 per million output tokens. After that promotional window, rates will rise to $3 per million input and $15 per million output tokens. This pricing structure positions Sonnet 5 as a cheaper alternative to heavyweight models like OpenAI’s GPT-5.5 and Google’s Gemini 3.1 Pro, though it remains pricier than Google’s lightweight Gemini 3.5 Flash.
How Sonnet 5 Compares on Benchmarks
The new model boasts notable performance gains over Sonnet 4.6, which debuted in February 2026. On agentic coding benchmarks, Sonnet 5 achieved a 63.2% score, compared to Sonnet 4.6’s 58.1% and the high-end Opus 4.8’s 69.2%. For knowledge-based tasks, Anthropic claims Sonnet 5 actually edge-cases past Opus 4.8, though the company still recommends Opus 4.8 for highly sensitive research requiring maximum accuracy. Developers looking to integrate these capabilities into their development pipelines can compare how this model stacks up against other specialized options in our guide to the best AI coding tools.
The Shift to Agentic Baselines
This release underscores a broader industry pivot from conversational chatbots to autonomous agents. TechCrunch reports that other major players are executing similar playbooks, pointing to OpenAI’s recent preview of GPT-5.6 Sol and Google’s Gemini 3.5 Flash launch in May. The primary battleground is no longer just raw capability, but how reliably and cheaply these models can perform tasks without human oversight.
Early testers of Sonnet 5 have noted its improved persistence and self-correction. A Zapier engineer cited by TechCrunch reported that the model successfully completed complex, multi-stage workflows—such as updating Salesforce records and emailing contacts—without stalling halfway through, a common issue with older model versions. Additionally, Anthropic claims the model is capable of checking its own output autonomously without being explicitly prompted to do so.
Frequently asked questions
How much does Claude Sonnet 5 cost?
Through August 31, 2026, Claude Sonnet 5 is priced at an introductory rate of $2 per million input tokens and $10 per million output tokens. After this date, the price will increase to $3 per million input tokens and $15 per million output tokens.
Is Claude Sonnet 5 available on free plans?
Yes, starting June 30, 2026, Claude Sonnet 5 is the default model for both free and Pro plans, and is available across all subscription tiers.
How does Sonnet 5 compare to Opus 4.8?
While Opus 4.8 remains the preferred choice for tasks requiring maximum accuracy, Sonnet 5 slightly outperforms it on certain knowledge work benchmarks and offers a much cheaper alternative, scoring 63.2% on agentic coding compared to Opus’s 69.2%.
To see how Anthropic’s coding performance measures up against top industry alternatives, read our guide to the best AI coding tools.
Best AI Coding Tools (2026): 7 Tested & Ranked →Source: TechCrunch. Published July 1, 2026.
Ali has hands-on tested 50+ AI tools and tracks model releases daily. Every verdict here comes from real, paid usage — never vendor demos or sponsored placements.
AI Tools Worth is independent and unsponsored. Some linked guides contain affiliate links — they never change our verdicts.