Anthropic Unveils Claude Sonnet 5, Offering Cost-effective AI Solutions

WHO:

Techmeme - Anthropic, a San Francisco-based AI lab, launched Claude Sonnet 5.

VentureBeat - Sualeh Asif, co-founder of Cursor, and Daniel Shepard, a senior engineer at Zapier, provided testimonials about Sonnet 5.

WHAT:

VentureBeat - Sonnet 5 scores 81.2% on OSWorld-Verified, up from Sonnet 4.6's score of 78.5%.

VentureBeat - Sonnet 5's introductory pricing is set at $2 per million input tokens and $10 per million output tokens through August 31, after which it will rise to $3 and $15 respectively, which is significantly lower than Opus 4.8's pricing of $5 per million input tokens and $25 per million output tokens.

VentureBeat - Sonnet 5 scores 63.2% on SWE-bench Pro, compared to Sonnet 4.6's score of 58.1%, and 80.4% on Terminal-Bench 2.1, up from Sonnet 4.6's 67.0%.

VentureBeat - On Humanity's Last Exam, Sonnet 5 scores 43.2% without tools and 57.4% with tools, while Opus 4.8 scores 57.9%.

VentureBeat - Sonnet 5 achieves a score of 1,618 on GDPval-AA v2, surpassing Opus 4.8's score of 1,615 and Sonnet 4.6's score of 1,395.

VentureBeat - Sonnet 5 is described as "the most agentic Sonnet model yet," capable of making plans and using tools like browsers and terminals.

Techmeme - Claude Sonnet 5 offers performance close to Opus 4.8 at a lower price and better agentic performance than Sonnet 4.6.

TechCrunch - Sonnet 5 is positioned as a cheaper alternative to Opus, GPT-5.5, and Gemini Pro.

VentureBeat - Sonnet 5 is the default model for users on Anthropic's Free and Pro plans and is available to Max, Team, and Enterprise customers.

VentureBeat - Introductory API pricing for Sonnet 5 is set at $2 per million input tokens and $10 per million output tokens through August 31, after which it will rise to $3 and $15 respectively.

VentureBeat - Sonnet 5 scores 63.2% on SWE-bench Pro, 80.4% on Terminal-Bench 2.1, 43.2% without tools and 57.4% with tools on Humanity's Last Exam, and 81.2% on OSWorld-Verified.

VentureBeat - Sonnet 5 scores 1,618 on GDPval-AA v2, surpassing Opus 4.8's score of 1,615.

WHERE:

VentureBeat - The launch is part of Anthropic's strategy as it approaches an initial public offering (IPO).

WHEN:

VentureBeat - The introductory pricing for Sonnet 5 is valid through August 31.

WHY:

VentureBeat - The emphasis on agentic capabilities reflects a shift in enterprise needs for AI systems that can operate with minimal human supervision, as enterprises are deploying AI systems that can navigate complex software environments and execute multi-step coding tasks.

VentureBeat - Anthropic aims to democratize access to powerful AI capabilities and drive broad-based developer adoption ahead of its IPO.

VentureBeat - The emphasis on agentic capabilities reflects a shift in enterprise needs for AI systems that can operate with minimal human supervision.

HOW:

VentureBeat - Sonnet 5 uses an updated tokenizer that changes how the model processes text, potentially increasing token counts by 1.0 to 1.35 times depending on content type.

VentureBeat - Sonnet 5 uses an updated tokenizer that changes how the model processes text, potentially increasing token counts by 1.0 to 1.35 times.

VentureBeat - Sonnet 5 includes cyber safeguards enabled by default to detect and block dangerous cybersecurity usage.

(UPDATED June 30, 2026 23:20 UTC)

Sonnet 5 has achieved an impressive score of 81.2% on OSWorld-Verified, a notable increase from Sonnet 4.6's score of 78.5%. The introductory pricing for Sonnet 5 is set at $2 per million input tokens and $10 per million output tokens through August 31, after which it will rise to $3 and $15 respectively, making it significantly cheaper than Opus 4.8. This new model features an updated tokenizer that enhances text processing, potentially increasing token counts by 1.0 to 1.35 times based on the type of content.

(UPDATED June 30, 2026 21:31 UTC)

Sonnet 5 has made significant strides in performance, scoring 63.2% on SWE-bench Pro and 80.4% on Terminal-Bench 2.1, both improvements over its predecessor, Sonnet 4.6. In addition, it achieved a score of 1,618 on GDPval-AA v2, surpassing Opus 4.8's score of 1,615, while also demonstrating its capabilities on Humanity's Last Exam with scores of 43.2% without tools and 57.4% with tools. This latest model is being hailed as "the most agentic Sonnet model yet," reflecting a growing demand for AI systems that can operate independently in complex environments, thanks in part to an updated tokenizer that enhances text processing.

(UPDATED June 30, 2026 19:43 UTC)

Anthropic has launched Claude Sonnet 5, a new AI model that offers performance close to Opus 4.8 at a lower price and improved agentic capabilities compared to Sonnet 4.6. The model will be the default for users on Anthropic's Free and Pro plans, with introductory API pricing set at $2 per million input tokens and $10 per million output tokens through August 31, after which prices will increase. With impressive scores on various benchmarks, Sonnet 5 aims to democratize access to powerful AI capabilities and meet the growing enterprise demand for systems that can operate with minimal human supervision.