Anthropic releases Claude Sonnet 5, says it is default AI model for free users
Anthropic has launched Claude Sonnet 5 and made it the default model for Free and Pro users. The company says it brings stronger autonomous task handling, lower pricing and tighter safety controls.
Just weeks after pulling the plug on public access to its most advanced AI models, Fable 5 and Mythos 5, following a US government national security and export control directive, Anthropic is back with a new model. This time, however, it is making the AI available for free to users across its markets. The company has unveiled Claude Sonnet 5, its newest AI model, and announced that it will become the default model for users on the Free and Pro plans.
The Claude Sonnet 5 model is now available across Anthropic's entire Claude lineup, including Max, Team and Enterprise subscriptions. Developers can also access it through Claude Code, the Claude API and the Claude Platform.
Claude Sonnet 5 features
Anthropic describes Claude Sonnet 5 as its most capable Sonnet model yet, with a strong focus on autonomous, or "agentic", AI. According to the company, the model can plan tasks, use tools such as browsers and terminals, and complete multi-step workflows with much less human intervention. It says these capabilities were previously limited to larger, more expensive models, but Sonnet 5 now brings them to a more affordable tier.
Anthropic also claims Sonnet 5 significantly improves on its predecessor, Claude Sonnet 4.6, across reasoning, coding, tool use and knowledge work. Internal benchmarks show gains on coding tests such as SWE-bench Pro and Terminal-Bench, as well as improvements in computer use and multidisciplinary reasoning. The company says Sonnet 5 narrows the performance gap with Claude Opus 4.8 while maintaining lower pricing.
The Claude Sonnet 5 is also said to be more efficient. Anthropic says Sonnet 5 offers a wider range of cost-performance options than Opus 4.8. Depending on the effort level selected, the model can deliver substantially better cost efficiency than Sonnet 4.6 and, at higher effort settings, match Opus 4.8 on certain tasks.
Claude Sonnet 5 price
To encourage adoption, Anthropic has introduced introductory API pricing of $2 per million input tokens and $10 per million output tokens until August 31, 2026. Standard pricing will then increase to $3 and $15 per million tokens, respectively.
Alongside the capability upgrades, Anthropic says it has strengthened the model's safety. According to the company, pre-deployment evaluations showed Claude Sonnet 5 is better than Sonnet 4.6 at refusing malicious requests, resisting prompt injection attacks, and reducing hallucinations and sycophancy. While it still trails Opus 4.8 on some alignment measures, Anthropic says Sonnet 5 is safer overall than its predecessor.
Meanwhile, Anthropic notes that Claude Sonnet 5 was not specifically trained for offensive cybersecurity tasks and launches with real-time cybersecurity safeguards enabled by default to detect and block dangerous use.
Meanwhile, Anthorpic says developers can start using Claude Sonnet 5 by simply switching from Claude Sonnet 4.6. The new model enables adaptive thinking by default, supports a 1 million-token context window and can generate up to 128,000 output tokens. It also uses a new tokenizer, meaning the same text is split into around 30 per cent more tokens than before. Claude Sonnet 5 is available through the Claude API, Amazon Bedrock, Google Cloud and Microsoft Foundry (in preview).
Just weeks after pulling the plug on public access to its most advanced AI models, Fable 5 and Mythos 5, following a US government national security and export control directive, Anthropic is back with a new model. This time, however, it is making the AI available for free to users across its markets. The company has unveiled Claude Sonnet 5, its newest AI model, and announced that it will become the default model for users on the Free and Pro plans.
The Claude Sonnet 5 model is now available across Anthropic's entire Claude lineup, including Max, Team and Enterprise subscriptions. Developers can also access it through Claude Code, the Claude API and the Claude Platform.
Claude Sonnet 5 features
Anthropic describes Claude Sonnet 5 as its most capable Sonnet model yet, with a strong focus on autonomous, or "agentic", AI. According to the company, the model can plan tasks, use tools such as browsers and terminals, and complete multi-step workflows with much less human intervention. It says these capabilities were previously limited to larger, more expensive models, but Sonnet 5 now brings them to a more affordable tier.
Anthropic also claims Sonnet 5 significantly improves on its predecessor, Claude Sonnet 4.6, across reasoning, coding, tool use and knowledge work. Internal benchmarks show gains on coding tests such as SWE-bench Pro and Terminal-Bench, as well as improvements in computer use and multidisciplinary reasoning. The company says Sonnet 5 narrows the performance gap with Claude Opus 4.8 while maintaining lower pricing.
The Claude Sonnet 5 is also said to be more efficient. Anthropic says Sonnet 5 offers a wider range of cost-performance options than Opus 4.8. Depending on the effort level selected, the model can deliver substantially better cost efficiency than Sonnet 4.6 and, at higher effort settings, match Opus 4.8 on certain tasks.
Claude Sonnet 5 price
To encourage adoption, Anthropic has introduced introductory API pricing of $2 per million input tokens and $10 per million output tokens until August 31, 2026. Standard pricing will then increase to $3 and $15 per million tokens, respectively.
Alongside the capability upgrades, Anthropic says it has strengthened the model's safety. According to the company, pre-deployment evaluations showed Claude Sonnet 5 is better than Sonnet 4.6 at refusing malicious requests, resisting prompt injection attacks, and reducing hallucinations and sycophancy. While it still trails Opus 4.8 on some alignment measures, Anthropic says Sonnet 5 is safer overall than its predecessor.
Meanwhile, Anthropic notes that Claude Sonnet 5 was not specifically trained for offensive cybersecurity tasks and launches with real-time cybersecurity safeguards enabled by default to detect and block dangerous use.
Meanwhile, Anthorpic says developers can start using Claude Sonnet 5 by simply switching from Claude Sonnet 4.6. The new model enables adaptive thinking by default, supports a 1 million-token context window and can generate up to 128,000 output tokens. It also uses a new tokenizer, meaning the same text is split into around 30 per cent more tokens than before. Claude Sonnet 5 is available through the Claude API, Amazon Bedrock, Google Cloud and Microsoft Foundry (in preview).
