Anthropic’s latest AI release, Claude Sonnet 5, crystallizes a new phase of the AI race: turning powerful “agents” into something cheap, safe, and ready for everyday work.
Early June 30: Anthropic sets the stage
On June 30, Anthropic unveiled Sonnet 5 as “the most agentic Sonnet model yet,” able to make plans, use browsers and terminals, and “run autonomously at a level that, just a few months ago, required larger and more expensive models.”1 The company said Sonnet 5 narrows the gap with its flagship Opus 4.8 “but at lower prices,” while safety tests show “a much lower ability to perform cybersecurity tasks than our current Opus models.”1
Anthropic also confirmed the model would be available immediately across all plans, with introductory pricing of $2 per million input tokens and $10 per million output tokens through August 31, before rising to $3 and $15.1
Media frame the launch: power vs. price vs. risk
Tech outlets quickly cast Sonnet 5 as a cost play. TechCrunch described it as “a more powerful and agentic version of the lab’s midsize model” and said its “pitch is confirmation that agentic capability is the new baseline expectation,” with differentiation now about “how cheaply” and reliably models can run autonomous work.2 The Next Web similarly called it “its most agentic mid-tier model yet,” noting it “runs close to the flagship Opus 4.8 on many tasks, but costs less than half as much.”3
Axios emphasized security and regulatory context. It reported Anthropic is releasing a “lower-priced model designed to bring more agentic AI capabilities to everyday users without the same cyber-risk profile as its most powerful systems,” and highlighted that Sonnet 5 “was not deliberately trained on cybersecurity tasks” and has a “much lower ability” to perform dangerous cyber activities than Opus.4 The outlet linked the launch to ongoing talks with the Trump administration, which had previously pushed Anthropic to take down its higher-powered Mythos and Fable models over security concerns.4
The Verge focused on the technology positioning Sonnet 5 as a “new mid-tier model, the successor to Sonnet 4.6,” quoting Anthropic’s claim that it can “run autonomously at a level that, just a few months ago, required larger and more expensive models,” with performance “close to that of Opus 4.8” and “a much lower ability to perform dangerous cybersecurity tasks than our current Opus models.”5
Later June 30: ecosystem reaction and deployment
As Sonnet 5 became the default Claude model for free and Pro tiers, third‑party platforms began integrating it. Anthropic’s own documentation noted that Sonnet 5 is available in Claude Code and through the Claude API for developers seeking a broader “range of cost-performance options.”1
Perplexity AI’s CEO Aravind Srinivas amplified the rollout, retweeting that “Claude Sonnet 5 is now available in Perplexity for Pro and Max subscribers” and can be selected “as an orchestrator model in Computer,”
6 and separately announcing: “New orchestrator model for Computer users (Pro and Max): Sonnet 5.”7
Across perspectives, the story is consistent: Sonnet 5 largely matches near‑flagship capability, is priced for aggressive agent deployment, and is deliberately constrained on offensive cyber skills—positioning Anthropic to grow agentic use while navigating mounting cost pressures and government scrutiny.