--- title: "Anthropic's Claude Sonnet 5 Matches Opus Performance at Half the Price" subtitle: "Claude Sonnet 5 launches as the new default on Free and Pro plans, matching Opus 4.8 performance at Sonnet pricing with stronger agentic autonomy" url: "https://alphasignal.ai/news/anthropic-s-claude-sonnet-5-matches-opus-performance-at-half-the-price" author: "AlphaSignal Newsroom" author_url: "https://alphasignal.ai/editorial-team" company: "Claude" company_url: "https://alphasignal.ai/company/claudeai" published: "2026-06-30T18:00:31.000Z" updated: "2026-06-30T18:05:28.998Z" type: "News" topics: - "Llms" - "Agents" subtopics: - "Long Context" - "Computer Use" - "Tool Use" upvotes: 51335 read_time: "5 min" source_url: "https://www.anthropic.com/news/claude-sonnet-5" comments: 0 --- # Anthropic's Claude Sonnet 5 Matches Opus Performance at Half the Price > Claude Sonnet 5 launches as the new default on Free and Pro plans, matching Opus 4.8 performance at Sonnet pricing with stronger agentic autonomy Anthropic just shipped [Claude Sonnet 5](https://www.anthropic.com/news/claude-sonnet-5), and the headline is hard to ignore: a mid-tier model that now performs close to Opus 4.8 across reasoning, coding, and agentic tasks, while keeping the same $3/$15 per million token pricing as its predecessor. It is live today across all Claude apps and the API, and it is the new default for Free and Pro users. The model is positioned squarely as an agentic workhorse. Anthropic describes it as their "most agentic Sonnet yet" , one that makes plans, uses tools like browsers and terminals, and runs autonomously at a level that, just a few months ago, required reaching for a larger and more expensive Opus-class model. ## The Agentic Leap That Actually Matters The core story here is not just benchmark numbers. It is a qualitative shift in how the model behaves during long-running tasks. Early access partners reported that Sonnet 5 finishes complex tasks where previous Sonnets stopped short, and crucially, it checks its own output without being asked. That kind of unprompted self-correction is the difference between a model that assists and one that actually gets work done. The agentic improvements show up in three concrete ways: - **Execution feedback loops:** Sonnet 5 uses a built-in terminal environment to run the code it writes, identify errors, and self-correct before presenting a solution. - **Multi-file contextual awareness:** The model maps out the entire dependency tree of a repository to ensure that a change in one file does not break a module three layers deep. - **Autonomous sub-agent coordination:** When faced with a massive feature request, the model partitions the task , one sub-agent writes the API endpoints while another generates the unit tests, all coordinated by the central Sonnet 5 agent. ![Benchmark comparison table showing Claude Sonnet 5 vs Sonnet 4.6 and Opus 4.8 across agentic coding, reasoning, computer use, and knowledge work](https://www-cdn.anthropic.com/images/4zrzovbb/website/9941d610909f28a504e16dd5af823df172ec6035-2600x1234.png) ## Near-Opus Performance at a Fraction of the Cost The benchmark picture is striking. The headline number is 92.4% on SWE-bench Verified , a benchmark that tests real GitHub issue resolution on novel codebases. For context, Claude Opus 4.6, their previous flagship, sat at 80.8%. SWE-bench Verified is a human-validated set of 500 real software engineering tasks, so this is not a synthetic test , it measures whether the model can actually ship a working patch. Computer use is the other standout. Sonnet 5 scores 88.3% on OSWorld-Verified. The human expert baseline on that benchmark is 72.4%, meaning Sonnet 5 is not just competitive with humans on desktop automation , it is meaningfully ahead. On reasoning: Sonnet 5 ships with a 2M token context window, now out of beta. And on developer preference: per Anthropic's internal numbers, developers preferred Sonnet 5 over Sonnet 4.6 in Claude Code roughly 82% of the time, citing fewer hallucinated completions, better cross-file context retention, and significantly improved frontend output quality. ## What It Is Built For Sonnet 5 is optimized for the kinds of tasks that benefit from sustained autonomy , not quick Q&A, but multi-step workflows where the model needs to plan, execute, and recover from errors without constant hand-holding. The sweet spots are: - Agentic coding pipelines in [Claude Code](https://claude.ai) or similar tools - Browser-based computer use and desktop automation - Long-horizon knowledge work: research synthesis, document analysis, financial workflows - Multi-agent orchestration where Sonnet 5 acts as a coordinator For tasks that are straightforward, single-turn, or where raw reasoning depth is the bottleneck, Claude Opus 4.8 remains the recommended starting point for the most complex tasks, as it is Anthropic's most capable Opus-tier model for complex reasoning, long-horizon agentic coding, and high-autonomy work. ## The Pricing Play Anthropic's decision to price Sonnet 5 at $3 per million input tokens and $15 per million output tokens is a direct shot at both Google and OpenAI. The model string is `claude-sonnet-5-20260401`, already live in claude.ai as the new default and on the API at the same $3/$15 per million tokens pricing as Sonnet 4.6. Introductory pricing is confirmed through August 31. Pricing for Sonnet 4.6 started at $3 per million input tokens and $15 per million output tokens, with up to 90% cost savings with prompt caching and 50% cost savings with batch processing , the same structure carries over to Sonnet 5. For teams running high-volume agentic workloads, those caching discounts make a meaningful difference at scale. ## Where to Get It Sonnet 5 is available right now across every surface: - Claude.ai (Free, Pro, Max, Team, Enterprise) , it is the new default for Free and Pro - The Anthropic API via model ID `claude-sonnet-5-20260401` - Amazon Bedrock, Google Cloud Vertex AI, Claude Platform on AWS, and Microsoft Foundry - Claude Code and the Claude iOS/Android apps ## The Bigger Picture Anthropic has been on a rapid cadence. Claude Opus 4.8 was announced on May 28, 2026 , roughly 42 days after Opus 4.7, the shortest gap so far between consecutive Claude Opus releases. Sonnet 5 arrives just weeks later, and the pattern is clear: Anthropic is compressing the performance gap between its mid-tier and flagship models with each generation. The assumption that needed updating after Sonnet 4.6 , that Opus-level agentic performance required Opus-level pricing , gets updated again here. Performance that would have previously required reaching for an Opus-class model is now available at Sonnet pricing. The question for teams running production agents is no longer "can we afford the capable model?" It is "what do we do with the compute budget we just freed up?" ![Bar chart comparing misaligned behavior scores across Sonnet 4.6, Mythos Preview, Opus 4.8, and Sonnet 5](https://www-cdn.anthropic.com/images/4zrzovbb/website/d018d76aa03c0ef18abc8a68de8f6fcd51c0a574-3840x2160.png) On the safety side, Anthropic also published alignment evaluations alongside the launch. The misaligned behavior scores , which measure things like deception, sycophancy, and power-seeking , show Sonnet 5 continuing the trend of improved alignment relative to previous Sonnet models, consistent with what Anthropic has emphasized across the 4.x generation. ## Sources - [X post by @claudeai](https://x.com/claudeai/status/2072017450611142835) - [Introducing Claude Sonnet 5 — Anthropic](https://www.anthropic.com/news/claude-sonnet-5) - [Anthropic just dropped Claude Sonnet 5, and the benchmarks are kind of insane — DEV Community](https://dev.to/best_codes/anthropic-just-dropped-claude-sonnet-5-and-the-benchmarks-are-kind-of-insane-3ppc) - [Claude Sonnet — Anthropic](https://www.anthropic.com/claude/sonnet) - [Models Overview — Claude Platform Docs](https://platform.claude.com/docs/en/about-claude/models/overview) - [Introducing Claude Sonnet 4.6 — Anthropic](https://www.anthropic.com/news/claude-sonnet-4-6)