--- title: Anthropic releases Claude Sonnet 5.5 with the cyber... description: Anthropic released Claude Sonnet 5.5, scoring 70.6% on the Terminal-Bench 4.0 agentic coding test, beating both its predecessor (10.3%) and the more expensive... canonical: https://daily.dev/posts/anthropic-releases-claude-sonnet-5-5-with-the-cyber-limits-it-reserved-for-its-best-models-ecrkmoc8s twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Anthropic releases Claude Sonnet 5.5 with the cyber limits it reserved for its best models | daily.dev og:description: Anthropic released Claude Sonnet 5.5, scoring 70.6% on the Terminal-Bench 4.0 agentic coding test, beating both its predecessor (10.3%) and the more expensive... og:url: https://daily.dev/posts/anthropic-releases-claude-sonnet-5-5-with-the-cyber-limits-it-reserved-for-its-best-models-ecrkmoc8s og:image: https://api.daily.dev/og/posts/eCrKmoc8s.png og:image:alt: Anthropic releases Claude Sonnet 5.5 with the cyber limits it reserved for its best models og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Anthropic releases Claude Sonnet 5.5 with the cyber limits it reserved for its best models **[The Next Web](https://daily.dev/sources/tnw)** · 3 min read · 0 upvotes · 0 comments ## Summary Anthropic released Claude Sonnet 5.5, scoring 70.6% on the Terminal-Bench 4.0 agentic coding test, beating both its predecessor (10.3%) and the more expensive flagship Opus 5.5 (66.4%). Priced at $2 per million input tokens and $10 per million output, it runs over 30% faster and costs up to 30% less per task than Sonnet 5. It's the first Sonnet to ship with frontier-style cyber safeguards, restricting higher-risk cyber requests to fall back to Sonnet 5, and the first with classifiers blocking reasoning extraction, following an August incident where researchers decoded 315,320 thinking blocks across OpenAI, Anthropic, and Google systems and recovered API keys and passwords. The EU AI Act's Article 55 requires such protections for general-purpose models with systemic risk, with the European Commission having gained fining power in August. Anthropic simultaneously claims the model doesn't advance capability frontiers while also calling its cyber improvements large enough to warrant frontier-level safeguards, and its pricing claim of parity with Sonnet 5 conflicts with earlier reporting of a scheduled price increase. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What does Claude Sonnet 5.5 score on Terminal-Bench 4.0 compared to Sonnet 5 and Opus 5.5? Claude Sonnet 5.5 scores 70.6% on the Terminal-Bench 4.0 agentic coding test, compared to 10.3% for Sonnet 5 and 66.4% for Opus 5.5 at its highest effort setting. Sonnet 5.5 also runs more than 30% faster and costs up to 30% less per task than its predecessor, despite Opus costing twice as much per token. _Track how model choice affects coding benchmarks and cost by following AI releases on daily.dev._ ### How much does Claude Sonnet 5.5 cost per million tokens? Claude Sonnet 5.5 is priced at $2 per million input tokens and $10 per million output tokens, which Anthropic describes as matching Sonnet 5's pricing. This conflicts with earlier reporting that those were introductory rates only until August 31, after which Sonnet 5 was to cost $3 input and $15 output, leaving the comparison basis unclear. _Developers budgeting API costs can follow pricing changes across model providers on daily.dev._ ### Why did Claude Sonnet 5.5 launch with cyber restrictions previously reserved for Opus-tier models? Anthropic found Sonnet 5.5's cybersecurity capabilities comparable to Opus 5, so it ships with frontier-style safeguards where higher-risk cyber requests fall back to Sonnet 5. It's also the first Sonnet with classifiers blocking reasoning extraction, following research that decoded 315,320 thinking blocks from public agent traces and recovered 62 API keys, 33 passwords, and seven private keys. _Anyone building on frontier models can track safety and safeguard changes affecting agent security on daily.dev._ ## Similar posts on daily.dev - [Anthropic Sonnet 5: It closes the gap with Opus 4.8, and is cheap until August](https://daily.dev/posts/anthropic-sonnet-5-it-closes-the-gap-with-opus-4-8-and-is-cheap-until-august-o0toggvty) · The New Stack · 0 upvotes · 0 comments - [Anthropic releases Sonnet 4.6](https://daily.dev/posts/anthropic-releases-sonnet-4-6-aufw9ozvi) · TechCrunch · 1 upvotes · 0 comments - [Announcing Claude Sonnet 5.5 on Snowflake Cortex AI](https://daily.dev/posts/announcing-claude-sonnet-5-5-on-snowflake-cortex-ai-lra59sbvf) · Snowflake · 2 upvotes · 0 comments --- Tags: [#ai-coding](https://daily.dev/tags/ai-coding), [#claude](https://daily.dev/tags/claude), [#anthropic](https://daily.dev/tags/anthropic), [#ai-safety](https://daily.dev/tags/ai-safety), [#ai-regulation](https://daily.dev/tags/ai-regulation) [View this post on daily.dev](https://daily.dev/posts/anthropic-releases-claude-sonnet-5-5-with-the-cyber-limits-it-reserved-for-its-best-models-ecrkmoc8s) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Anthropic releases Claude Sonnet 5.5 with the cyber limits it reserved for its best models","url":"https://daily.dev/posts/anthropic-releases-claude-sonnet-5-5-with-the-cyber-limits-it-reserved-for-its-best-models-ecrkmoc8s","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/anthropic-releases-claude-sonnet-5-5-with-the-cyber-limits-it-reserved-for-its-best-models-ecrkmoc8s"},"datePublished":"2026-09-28T18:43:14.835Z","dateModified":"2026-09-28T19:41:15.217Z","description":"Anthropic released Claude Sonnet 5.5, scoring 70.6% on the Terminal-Bench 4.0 agentic coding test, beating both its predecessor (10.3%) and the more expensive...","image":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/997a6fafbbbe1e5fd54e847c3e6d12d2?_a=AQAEuop","thumbnailUrl":"https://media.daily.dev/image/upload/f_auto,q_auto/v1/posts/997a6fafbbbe1e5fd54e847c3e6d12d2?_a=AQAEuop","isAccessibleForFree":true,"articleSection":"The Next Web","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"The Next Web","logo":"https://media.daily.dev/image/upload/t_logo,f_auto/v1/logos/tnw","url":"https://daily.dev/sources/tnw"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/anthropic-releases-claude-sonnet-5-5-with-the-cyber-limits-it-reserved-for-its-best-models-ecrkmoc8s","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-coding,claude,anthropic,ai-safety,ai-regulation","timeRequired":"PT3M"} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"The Next Web","item":"https://daily.dev/sources/tnw"},{"@type":"ListItem","position":3,"name":"Anthropic releases Claude Sonnet 5.5 with the cyber limits it reserved for its best models"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/anthropic-releases-claude-sonnet-5-5-with-the-cyber-limits-it-reserved-for-its-best-models-ecrkmoc8s#faq","mainEntity":[{"@type":"Question","name":"What does Claude Sonnet 5.5 score on Terminal-Bench 4.0 compared to Sonnet 5 and Opus 5.5?","acceptedAnswer":{"@type":"Answer","text":"Claude Sonnet 5.5 scores 70.6% on the Terminal-Bench 4.0 agentic coding test, compared to 10.3% for Sonnet 5 and 66.4% for Opus 5.5 at its highest effort setting. Sonnet 5.5 also runs more than 30% faster and costs up to 30% less per task than its predecessor, despite Opus costing twice as much per token. Track how model choice affects coding benchmarks and cost by following AI releases on daily.dev."}},{"@type":"Question","name":"How much does Claude Sonnet 5.5 cost per million tokens?","acceptedAnswer":{"@type":"Answer","text":"Claude Sonnet 5.5 is priced at $2 per million input tokens and $10 per million output tokens, which Anthropic describes as matching Sonnet 5's pricing. This conflicts with earlier reporting that those were introductory rates only until August 31, after which Sonnet 5 was to cost $3 input and $15 output, leaving the comparison basis unclear. Developers budgeting API costs can follow pricing changes across model providers on daily.dev."}},{"@type":"Question","name":"Why did Claude Sonnet 5.5 launch with cyber restrictions previously reserved for Opus-tier models?","acceptedAnswer":{"@type":"Answer","text":"Anthropic found Sonnet 5.5's cybersecurity capabilities comparable to Opus 5, so it ships with frontier-style safeguards where higher-risk cyber requests fall back to Sonnet 5. It's also the first Sonnet with classifiers blocking reasoning extraction, following research that decoded 315,320 thinking blocks from public agent traces and recovered 62 API keys, 33 passwords, and seven private keys. Anyone building on frontier models can track safety and safeguard changes affecting agent security on daily.dev."}}]} ```