\n<\/p>\n
As shipping agentic capabilities becomes table stakes among foundation model companies, Anthropic is releasing Claude Sonnet 5, a more powerful and agentic version of the lab\u2019s midsize model.\u00a0<\/p>\n
\u201cIt 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,\u201d Anthropic said in a blog post<\/a>.\u00a0<\/p>\n That framing mirrors what OpenAI and Google have said about their own recent releases. OpenAI\u2019s GPT-5.6 Sol<\/a> was launched in preview last week, and it is also the firm\u2019s most agentic model yet, allowing users to split work across subagents for longer autonomous tasks. Google\u2019s Gemini 3.5 Flash<\/a>, which launched in May, was pitched as a shift from a conversational chatbot to an agentic tool that plans, builds, and iterates on real work with minimal human input.<\/p>\n Sonnet 5\u2019s pitch is confirmation that agentic capability is the new baseline expectation at every price tier. Now the differentiator isn\u2019t going to be who can do agentic work best, but how cheaply they can do it and how reliably without human oversight.\u00a0\u00a0<\/p>\n Sonnet 5 promises performance close to that of Opus 4.8<\/a>, but for much lower costs. Starting Tuesday, Claude Sonnet 5 will be the default model for free and Pro plans and is available for every subscription.<\/p>\n At launch, Sonnet 5 is priced at $2 per million input tokens and $10 per million output tokens through August 31, after which the price will jump to $3 per million input tokens and $15 per million output tokens. That makes Sonnet 5 cheaper than Opus 4.8, as well as OpenAI\u2019s GPT-5.5 and Google\u2019s Gemini 3.1 Pro. (It\u2019s still more expensive than Gemini 3.5 Flash.)<\/p>\n The new model also demonstrates significant improvements over its predecessor Sonnet 4.6, released in February<\/a>, on agentic performance like reasoning, tool use, software coding, and knowledge work, according to Anthropic.\u00a0<\/p>\n For example, on one benchmark, Sonnet 5 scores a 63.2% on agentic coding, compared to Opus 4.8\u2019s 69.2% and Sonnet 4.6\u2019s 58.1%. On a knowledge work benchmark, Sonnet 5 actually slightly outperforms Opus 4.8, which is known for winning on solving the hardest problems like making subtle judgment calls and deep research.\u00a0<\/p>\n \u201cOpus 4.8 is still the model of choice for higher accuracy on these tasks, but Sonnet 5 provides developers with lower-priced options that are of much higher quality than what was previously available,\u201d Anthropic says. \u201cBetween Sonnet 5 and Opus 4.8, users can adjust the effort level to find the right balance of cost and performance.\u201d<\/p>\n According to testers cited in the blog post, Sonnet 5 also excels at finishing complex tasks where previous model versions would have stopped short and \u201cchecks its own output without explicitly being asked.\u201d<\/p>\n \u201cWe handed Claude Sonnet 5 a two-part job \u2014 update Salesforce account tiers, send a launch announcement to enterprise contacts \u2014 and it finished end to end,\u201d Daniel Shepard, a senior engineer at Zapier, said in a statement. \u201cThat used to stall halfway. For day-to-day automation, it\u2019s a no-brainer. \u201d<\/p>\n On safety, Sonnet 5 also demonstrates a lower rate of \u201cundesirable behaviors\u201d like cooperation with misuse and deception than its predecessor, making it safer to use in agentic contexts. It\u2019s better at refusing malicious requests and sidestepping hijack attempts in prompt-injection attacks. It also hallucinates and engages in sycophantic behavior at a lower rate than Sonnet 4.6.<\/p>\n That said, it\u2019s not on the same level as Opus 4.8 and Claude Mythos Preview when it comes to misaligned behavior. \u201cEvaluations also show that it has a much lower ability to perform dangerous cybersecurity tasks than our current Opus models,\u201d reads the blog post.<\/p>\n Lovable co-founder Fabian Hedin said in a statement that Claude Sonnet 5 \u201crefuses unsafe requests cleanly and consistently.\u201d<\/p>\n \u201cAt Lovable, we\u2019re putting powerful tools in the hands of millions of builders,\u201d Hedin said. \u201cA model that knows when to say no is just as important as one that knows how to build.\u201d<\/p>\n Updated to correct that the price of output tokens is $15 per million output tokens after August 31.<\/em><\/p>\n<\/div>\n When you purchase through links in our articles, we may earn a small commission<\/a>. This doesn\u2019t affect our editorial independence.<\/em><\/p>\n