发布 Claude Sonnet 5：面向编程、智能代理与规模化专业工作的前沿性能
英文标题：ProductJun 30, 2026Introducing Claude Sonnet 5Sonnet 5 delivers frontier performance across coding, agents, and professional work at scale.
Anthropic 发布 Claude Sonnet 5，是迄今为止最擅自主代理任务的 Sonnet 模型。其代理性能大幅逼近 Opus 4.8，但价格更低，在推理、工具使用、编程和知识工作等方面较 Sonnet 4.6 显著提升。即日起在所有套餐中可用，API 引入定价为每百万输入令牌 2 美元、输出令牌 10 美元，持续至 2026 年 8 月 31 日。该模型在安全评估中表现出更低的不良行为率，并将默认启用网络安全防护，对开发者与企业意味着更高效、更安心的自主代理部署。
96.5 KB
49.7 KB
83.5 KB
85.3 KB
7.9 KB
5.1 KB
9.6 KB
2.7 KB
3.7 KB
3.4 KB
9.7 KB
2.9 KB
9.0 KB
4.1 KB
1.8 KB
4.7 KB
5.7 KB
41.7 KB
57.0 KB
3.2 MB
中文
Claude Sonnet 5 旨在成为有史以来最自主代理的 Sonnet 模型。它可以制定计划，使用浏览器和终端等工具，并能在过去几个月仍需更大型、更昂贵模型才能实现的水平上自主运行。
对许多开发者而言，代理式 AI 时代始于 Sonnet 系列模型：Claude Sonnet 3.5、3.6 和 3.7 是最早展现出卓越编码与工具使用技能的模型。然而最近，代理能力最显著的提升一直集中在我们的 Opus 级模型。
Sonnet 5 缩小了这一差距：其性能接近 Opus 4.8，但价格更低。它在推理、工具使用、编码和知识工作等代理性能的重要方面，相比前代 Sonnet 4.6 有了大幅提升：
Sonnet 5 在多项评估中的得分，与 Sonnet 4.6 和 Opus 4.8（均为更综合的模型，供参考）的对比。Claude Sonnet 5 系统卡 中详细报告了更广泛的评估。
我们的安全评估发现，Sonnet 5 的不良行为总体比例低于 Sonnet 4.6，并且在代理式环境中使用通常更为安全。评估还表明，它执行网络安全任务的能力远低于我们当前的 Opus 模型。
即日起，Claude Sonnet 5 在所有套餐中可用：它是 Free 和 Pro 套餐的默认模型，Max、Team 和 Enterprise 用户亦可使用。它也在 Claude Code 和 Claude 平台上可用，并以引入定价上线：每百万输入令牌 2 美元，每百万输出令牌 10 美元，有效期至 2026 年 8 月 31 日，之后将调整为每百万输入令牌 3 美元、输出令牌 15 美元。开发者可通过 Claude API 使用 claude-sonnet-5。
与 Claude Sonnet 5 协同工作
下面的图表比较了 Sonnet 5、Sonnet 4.6 和 Opus 4.8 在不同努力程度下，于代理式搜索评估 BrowseComp 和计算机使用评估 OSWorld-Verified 中的性能表现。Sonnet 5（橙色线）严格优于 Sonnet 4.6（灰色线），并覆盖了比 Opus 4.8（黄色线）更广泛的成本-性能选择区间。它在中低努力程度下显著提升了成本效率；其高努力程度的性能在某些任务上可与 Opus 4.8 匹敌。在 Sonnet 5 和 Opus 4.8 之间，用户可以调整努力程度，以找到成本与性能的最佳平衡。
我们早期试用伙伴的反馈一致表明：Sonnet 5 比其前代更具代理性。测试人员描述了它如何完成以往 Sonnet 模型会半途而废的复杂任务，如何在未被明确要求的情况下检查自己的输出，以及如何以极具吸引力的价格完成所有这些代理工作：
Claude Sonnet 5 为我们的代理提供了执行多层软件工程任务的强大执行层。它能很好地处理持续的编码、工具使用和调试工作，适应杂乱的技术环境，并且在需要持续跟进和技术根基的工作流中尤其有用。
我们交给 Claude Sonnet 5 一个由两部分组成的任务——更新 Salesforce 客户层级，并向企业联系人发送发布通知——它端到端地完成了。以前通常会半路卡住。对于日常自动化，这是无需犹豫的选择。
Claude Sonnet 5 用更少的资源完成更多任务。输出质量不变，需要的步骤更少。它还能清晰、一致地拒绝不安全请求。在 Lovable，我们正在为百万建设者提供强大工具。一个懂得何时说“不”的模型，与懂得如何构建的模型同样重要。
我们用 Claude Sonnet 5 跑了数十个最具挑战性的真实拉取请求，它独自完成了每一个，一直到测试和验证结果——使我们的工程师能够专注于判断、决策和最终签核。
我让 Claude Sonnet 5 调查一个 bug。在没有提示的情况下，它编写了一个复现测试，实现了修复，然后暂存修复以确认在没有更改的情况下 bug 会重现。所有操作一气呵成。
有了 Claude Sonnet 5，代理能够按计划执行，遵循我们的惯例，并以高效成本交付清晰的多步更改。
Claude Sonnet 5 在处理棕地代码方面表现最佳——竞态条件、隐藏的测试、没人愿意碰的部分。它能追溯故障的根本原因，并交付稳定修复，而不是治标不治本。
Claude Sonnet 5 位于 Eve 的原告法律工作的帕累托前沿。在法律研究和分析方面，我们看到了最明显的提升，其性价比让迁移决策变得容易。
ClickHouse 代理探索实时数据并即时生成洞察，因此在测试新模型时，获取洞察的时间至关重要。Claude Sonnet 5 用更紧凑的步骤推理，能明显更快地让用户得到答案。这种速度是我们的客户能感受到的差异。
在 Pace，我们的计算机使用代理在运营团队已使用的系统上运行保险工作流——投保接收、出险首报、损失记录等。Claude Sonnet 5 始终采取正确的行动，并且快速执行，这正是真实保险业务所需。
对于管理大流量、复杂工作负载的企业团队而言，Claude Sonnet 5 代表着真正的进步——在关键领域性能强劲，且速度和成本组合使规模化部署变得实际可行。在几项复杂任务上，它超过了当前前沿水平，同时以低成本提供快速响应。对于大规模运行的企业来说，这是一个真正的运营胜利。
Claude Sonnet 5 处理了我们测试过的全范围编码任务，同时解决了更多问题。这是质量和效率的双重实质性提升。
Claude Sonnet 5 是一个强大的代理编码模型，提供与 Opus 级模型相当的顶级准确性，并且相比 Sonnet 4.6 有明显的跃迁式改进。它探索透彻，并在复杂任务上可明显更持久地保持专注。
安全评估
我们部署前的安全评估发现，Sonnet 5 总体上较 Sonnet 4.6 有所改善。在代理安全方面，该模型在拒绝恶意请求和抵御提示注入攻击方面表现更佳。与 Sonnet 4.6 相比，该模型的幻觉和迎合行为比率更低。在我们的自动化行为审计（测试范围广泛的不对齐行为，如配合滥用和欺骗）中，Sonnet 5 的整体得分较低（即更安全）。然而，在这项评估中，与能力更强的 Opus 4.8 和 Claude Mythos Preview 相比，它确实显示出略高的不对齐行为发生率。
我们自动化行为审计中的不对齐行为率，该审计测试了多种情景和上下文中的广泛不良行为（完整列表和各行为的详细结果见Sonnet 5 系统卡第 6.4 节）。Sonnet 5 的不对齐行为率总体低于 Sonnet 4.6，但高于 Mythos Preview 和 Opus 4.8。
我们没有刻意针对网络安全任务训练 Sonnet 5。它可以执行一些常规的、无害的网络任务，但在测试潜在危险网络技能（如开发软件漏洞利用）的评估中，其表现远逊于 Opus 4.8 和 Mythos 5 等模型。下方图表展示了一项评估的结果，该评估测试了模型针对 Firefox 浏览器漏洞开发漏洞利用的能力。Sonnet 5 从未成功开发出完整的可用漏洞利用，但其_部分_成功率略高于 Sonnet 4.6。后一变化很可能源于通用智能的提升，而非特定训练。
衡量模型针对 Firefox 147 中软件漏洞开发漏洞利用成功率的分数（此评估与 Mozilla 合作开发；所有漏洞已在 Firefox 148 中修补）。对于每个模型，左侧条形表示模型（无安全防护）成功开发可用漏洞利用的次数百分比；右侧条形表示模型取得部分成功的次数百分比。两款 Sonnet 模型均未能成功开发出可用漏洞利用（分数均为 0.0%）；Sonnet 5 的部分成功率略高于 Sonnet 4.6。两款 Sonnet 模型的网络能力均远逊于 Opus 4.8 和 Mythos 5。完整详情见 Sonnet 5 系统卡第 3.2.4 节。
由于 Sonnet 5 在这些任务上比前代稍强，我们默认启用了网络安全防护。这些防护措施——可实时检测和阻断危险的网络使用——与 Claude Opus 4.7 和 4.8 中已有的防护相同（因为我们判定 Sonnet 5 的整体网络安全风险水平较低，这些防护不如与 Fable 5 一起推出的防护那样严格，后者会阻断范围更广的网络安全任务）。1
我们针对 Sonnet 5 在众多安全和能力评估中的全面评估，报告于 Claude Sonnet 5 系统卡中。
可用性与定价
Claude Sonnet 5 今日起在所有地区可用，引入定价为每百万输入令牌 2 美元、每百万输出令牌 10 美元，有效期至 2026 年 8 月 31 日。之后将切换到标准定价，即每百万输入令牌 3 美元、输出令牌 15 美元。2 我们已提高 Chat、Cowork、Claude Code 和 Claude 平台3 的速率限制，以适应更高努力程度带来的更高令牌消耗；用户可根据自己的项目需要选择合适的努力等级。
变更日志
编辑于 2026 年 6 月 30 日：在此文章的原始版本中，我们包含了一张基于较简单方法得出的 BrowseComp 评估成本-性能图表，该方法未能反映我们用于代理搜索评估的标准方法。这导致低估了 Sonnet 5 在该评估中的表现。
我们现已更新该图表，使其符合我们在Sonnet 5 系统卡中使用和讨论的方法（采用 1000 万令牌预算，结合压缩和程序化工具调用）。我们还更新了相关文本。
English / 原文
Claude Sonnet 5 is built to be the most agentic Sonnet model yet. 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.
For many developers, the agentic AI era began with Sonnet-class models: Claude Sonnet 3.5, 3.6, and 3.7 were the first models that showed impressive skills in coding and tool use. More recently, though, the clearest gains in agentic capabilities have been in our Opus-class models.
Sonnet 5 narrows the gap: its performance is close to that of Opus 4.8, but at lower prices. It’s a substantial improvement over its predecessor, Sonnet 4.6, on important aspects of agentic performance like reasoning, tool use, coding, and knowledge work:
Scores for Sonnet 5 on a variety of evaluations compared to those of Sonnet 4.6 and Opus 4.8 (a more generally capable model, for reference). TheClaude Sonnet 5 System Card reports a broader set of evaluations in detail.
Our safety assessments found that Sonnet 5 shows an overall lower rate of undesirable behaviors than Sonnet 4.6, and is generally safer to use in agentic contexts. Evaluations also show that it has a much lower ability to perform cybersecurity tasks than our current Opus models.
From today, Claude Sonnet 5 is available across all plans: it is the default model for Free and Pro plans, and is available to Max, Team, and Enterprise users. It’s also available in Claude Code and on the Claude Platform, where it launches with introductory pricing of $2 per million input tokens and $10 per million output tokens through August 31, 2026, after which it will be priced at $3 per million input tokens and $15 per million output tokens. Developers can use claude-sonnet-5 via the Claude API.
Working with Claude Sonnet 5
The charts below compare the performance of Sonnet 5 with Sonnet 4.6 and Opus 4.8 at different effort levels on the agentic search evaluation BrowseComp and the computer use evaluation OSWorld-Verified. Sonnet 5 (orange line) is a strict improvement over Sonnet 4.6 (gray line) and covers a much wider range of cost-performance options than Opus 4.8 (yellow line). It provides substantially improved cost efficiency at medium effort; its higher-effort performance can match Opus 4.8 on some tasks. Between Sonnet 5 and Opus 4.8, users can adjust the effort level to find the right balance of cost and performance.
Feedback from our early access partners has been consistent: Sonnet 5 is much more agentic than its predecessors. Testers described how it finishes complex tasks where previous Sonnet models would stop short, how it checks its own output without explicitly being asked, and how it does all this agentic work at an attractive price point:
Claude Sonnet 5 gives our agents a strong execution layer for multi-step software engineering work. It handles sustained coding, tool use, and debugging well across messy technical contexts, and has been especially useful for workflows where follow-through and technical grounding matter.
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. That used to stall halfway. For day-to-day automation, it’s a no-brainer.
Claude Sonnet 5 gets more done with less. Same output quality, fewer steps to get there. It refuses unsafe requests cleanly and consistently, too. At Lovable, we’re putting powerful tools in the hands of millions of builders. A model that knows when to say no is just as important as one that knows how to build.
We ran Claude Sonnet 5 against dozens of our most challenging real pull requests, and it carried each one through to a tested, verified result on its own — freeing our engineers to focus on the judgment, the decision, and the final sign-off.
I asked Claude Sonnet 5 to investigate a bug. Unprompted, it wrote a reproducing test, implemented the fix, then stashed it to confirm the bug came back without the change. All in a single pass.
With Claude Sonnet 5, agents stay on plan, follow our conventions, and ship clean multi-step changes, all at an efficient cost.
Claude Sonnet 5 is at its best on brownfield code—race conditions, hidden tests, the parts nobody wants to touch. It traces a failure to its actual root cause and ships a durable fix instead of patching the symptom.
Claude Sonnet 5 sits on the Pareto frontier for Eve’s plaintiff-law tasks. We see the clearest gains in legal research and analysis, at a price-to-performance ratio that made the choice to migrate easy.
ClickHouse agents explore live data and produce insights on the fly, so time-to-insight matters when testing new models. Claude Sonnet 5 reasons in tighter steps and gets our users to answers noticeably faster. That speed is a difference our customers feel.
At Pace, our computer-use agents run insurance workflows—submission intake, FNOL, loss runs—on the systems our operations teams already use. Claude Sonnet 5 consistently takes the right action and does it quickly, which is what real insurance work demands.
For enterprise teams managing high-volume, complex workloads, Claude Sonnet 5 represents a genuine step forward — strong performance where it counts, with the speed and cost profile that makes scaling practical. On several complex tasks it exceeds the current frontier, while delivering fast responses at low costs. For enterprises running at scale, that's a real operational win.
Claude Sonnet 5 handled the full range of coding tasks we tested it on, while resolving more issues. It's a meaningful improvement to both quality and efficiency.
Claude Sonnet 5 is a strong agentic-coding model, delivering top-tier accuracy comparable to Opus-class models and a clear step-function improvement over Sonnet 4.6. It conducts thorough explorations and sustains focus noticeably longer on complex tasks.
Safety evaluations
Our pre-deployment safety evaluations found that Sonnet 5 was overall an improvement on Sonnet 4.6. On agentic safety, the model is better at refusing malicious requests and resisting hijack attempts in prompt injection attacks. The model shows lower rates of hallucination and sycophancy than Sonnet 4.6. On our automated behavioral audit, which tests a wide range of misaligned behaviors such as cooperation with misuse and deception, Sonnet 5 scored lower (that is, safer) overall. However, it did show somewhat higher rates of misaligned behavior on this assessment compared to the more capable Opus 4.8 and Claude Mythos Preview.
Rates of misaligned behavior on our automated behavioral audit, which tests for a very wide range of undesirable behaviors across many situations and contexts (see Section 6.4 of theSonnet 5 System Card for a complete list and results for each specific behavior). Sonnet 5 shows an overall lower rate of misaligned behavior than Sonnet 4.6, though a higher rate than Mythos Preview and Opus 4.8.
We did not deliberately train Sonnet 5 on cybersecurity tasks. It can perform some routine, non-harmful cyber tasks, but on evaluations testing potentially dangerous cyber skills, such as developing software exploits, it shows substantially poorer performance than models such as Opus 4.8 and Mythos 5. Scores from one evaluation, which tested models’ ability to develop exploits for vulnerabilities in the Firefox browser, are shown in the chart below. Sonnet 5 was never able to develop a full working exploit, but it does show a slightly higher rate of partial success than Sonnet 4.6. This latter change is likely due to improvements in general intelligence rather than specific training.
Scores measuring models’ success at developing exploits for software vulnerabilities in Firefox 147 (this evaluation was developedin collaboration with Mozilla; all vulnerabilities have been patched in Firefox 148). For each model, the left-hand bar shows how often the model (without safeguards) developed a working exploit; the right-hand bar shows how often the model had partial success. Neither of the Sonnet models could successfully develop a working exploit (both scored 0.0%); Sonnet 5 showed a slightly higher partial success rate than Sonnet 4.6. Both Sonnet models have substantially poorer cyber capabilities than Opus 4.8 and Mythos 5. For full details, see Section 3.2.4 of the Sonnet 5 System Card.
Since Sonnet 5 is somewhat stronger than its predecessor on these tasks, we’ve launched it with cyber safeguards enabled by default. These safeguards—which detect and block dangerous cyber usage in real time—are the same as those present in Claude Opus 4.7 and 4.8 (because we judged that the overall level of cybersecurity risk from Sonnet 5 was low, the safeguards are less strict than those launched with Fable 5, which block a much wider range of cybersecurity tasks).1
Our full assessment of Sonnet 5 across many safety and capability evaluations is reported in the Claude Sonnet 5 System Card.
Availability and pricing
Claude Sonnet 5 is available everywhere today at an introductory price of $2 per million input tokens and $10 per million output tokens through August 31, 2026. It then moves to standard pricing at $3 per million input tokens and $15 per million output tokens.2 We’ve increased rate limits across Chat, Cowork, Claude Code, and the Claude Platform3 to accommodate the higher token usage of higher effort levels; users can select whichever level makes sense for their particular project.
Changelog
Edit June 30, 2026: In the original version of this post, we included a cost-performance chart for the BrowseComp evaluation that was based on data from a simpler methodology that did not reflect thestandard methodology we use for agentic search evaluations. This had the result of underestimating Sonnet 5's performance on the evaluation.
We have now updated the chart so that it matches the methodology that we used and discussed in theSonnet 5 system card (which used a 10M token budget with compaction and programmatic tool calling). We have also updated the surrounding text.