Anthropic released Claude Sonnet 5 on Tuesday, a midsize model that narrows the performance gap with Opus 4.8 on agentic tasks while costing significantly less to run. Sonnet 5 is now the default model for free and Pro plan users across Claude.ai, Claude Code, and the Claude API.
The pricing undercuts every competing frontier model except Google’s Gemini 3.5 Flash. Sonnet 5 launches at $2 per million input tokens and $10 per million output tokens through August 31, 2026, after which it rises to $3 and $15 respectively. For comparison, Opus 4.8 runs $5/$25, and OpenAI’s GPT-5.5 and Google’s Gemini 3.1 Pro are both priced higher, according to TechCrunch.
Benchmark Performance
On agentic coding benchmarks, Sonnet 5 scores 63.2% compared to Opus 4.8’s 69.2% and its predecessor Sonnet 4.6’s 58.1%. On knowledge work benchmarks, Sonnet 5 slightly outperforms Opus 4.8, according to Anthropic’s blog post.
“It 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,” Anthropic wrote.
The model also demonstrates improved safety metrics. Sonnet 5 shows lower rates of hallucination, sycophancy, and “undesirable behaviors” like cooperation with misuse compared to Sonnet 4.6. It refuses malicious requests more consistently and resists prompt injection attacks more effectively, according to Anthropic’s safety evaluations.
Early Deployment Results
Zapier senior engineer Daniel Shepard described a concrete workflow improvement: “We handed Claude Sonnet 5 a two-part job, update Salesforce account tiers, send a launch announcement to enterprise contacts, and it finished end to end,” he said in a statement to TechCrunch. “That used to stall halfway.”
Lovable co-founder Fabian Hedin highlighted the safety angle: “A model that knows when to say no is just as important as one that knows how to build,” he told TechCrunch.
The Pricing Pressure
Sonnet 5 arrives alongside OpenAI’s GPT-5.6 Sol (launched last week in limited US-only preview) and Google’s Gemini 3.5 Flash (May). All three position agentic capabilities as table stakes rather than premium features.
The competitive implication is straightforward: if near-Opus agentic performance is available at Sonnet pricing, the cost barrier for running autonomous agents in production drops substantially. Developers who previously reserved Opus-class models for complex agent tasks can now run similar workloads at roughly 40% of the cost. Anthropic explicitly encourages this tradeoff, noting that users “can adjust the effort level to find the right balance of cost and performance” between Sonnet 5 and Opus 4.8.