The Developer Habits Report

"The change sweeping through software development is astounding. This inaugural Developer Habits Report, based on Cursor data, captures that transformation across five themes."

"软件开发正在经历的变化令人震惊。这份首发的《开发者习惯报告》基于 Cursor 数据,用五个主题记录了这场转型。"

"Developer acceleration. We chart how coding speed has doubled year-over-year, PRs are getting larger and deeper, and agent-generated code is surviving review at higher rates than ever."

"开发者加速。我们记录了写代码的速度如何一年翻倍、PR 如何越来越大越来越深,以及 agent 生成的代码以前所未有的比例存活下来、通过审查。"

"The economics of intelligence. We benchmark seven model families on cost per line and cost per submit, revealing wide heterogeneity in unit economics."

"智能的经济学。我们对七个模型家族在"每行成本"和"每次提交成本"上做了基准测试,揭示出单位经济模型之间的巨大差异。"

"The power user gap. We find that while AI is leading to broad productivity gains, the change is most pronounced in the top 1% of developers."

"超级用户鸿沟。我们发现,虽然 AI 带来了普遍的生产力提升,但这种变化在头部 1% 的开发者身上最为显著。"

"The rise of context. We show the dramatic increase in input tokens, and the shift toward cache-read tokens, which is giving agents the working memory to take on more complex tasks and produce higher-quality code."

"上下文的崛起。我们展示了输入 token 的急剧增长,以及向 cache-read token 的转移——正是这些,给了 agent 承担更复杂任务、产出更高质量代码所需的"工作记忆"。"

"The shift to automation. Finally, we look at how coding agents are evolving from a tool used by individual developers into an entire system for building and maintaining software, often automatically."

"转向自动化。最后,我们考察了编码 agent 如何从"个体开发者使用的工具",演化成一整套构建和维护软件的系统——而且常常是自动完成的。"

"This report provides a data-driven fixed point for understanding where agentic software development stands today, and where it appears to be headed next."

"这份报告提供了一个由数据锚定的坐标点,帮我们看清 agentic 软件开发今天走到了哪里,以及它接下来正朝哪个方向去。"

"Developers are working faster and producing more code, but the changes in developer productivity go far beyond volume."

"开发者干得更快、产出的代码更多,但生产力的变化远不止"量"这一点。"

"AI is also changing the shape of the work, with PRs getting larger, agent sessions going deeper, and AI-generated code surviving in codebases for longer periods of time."

"AI 还在改变工作的"形状":PR 越来越大,agent 会话越来越深,AI 生成的代码在代码库里存活得越来越久。"

"Developers are adding more code per week, with growth accelerating since the start of 2026. While this is not a perfect metric, it provides a directionally interesting baseline for understanding how developer work is changing."

"开发者每周新增的代码越来越多,自 2026 年初起增速加快。这虽然不是一个完美的指标,但提供了一个方向性上很有意思的基线,帮我们理解开发者的工作正在如何变化。"

"Lines added per PR are up roughly 2.5x year over year and the growth rate is accelerating."

"单个 PR 的代码增量同比上涨约 2.5 倍,而且增速还在加快。"

"Mega PRs, defined as PRs with at least 1,000 lines changed, are becoming more common as developers use AI to take on larger units of work in a single PR. It's interesting to see the jump in mega PRs in January 2026, when many developers were trying out the latest improvements in coding agents and models."

"巨型 PR(定义为改动至少 1,000 行的 PR)正变得越来越常见——开发者在用 AI 把更大块的工作塞进单个 PR 里。有意思的是,2026 年 1 月巨型 PR 出现了一次跳升,当时很多开发者正在试用编码 agent 和模型的最新改进。"

"In just the last two months, average tool calls per session have risen roughly 30%. Coding agents are taking on more complex work, reading and editing files, searching code, running shell commands, and browsing the web more frequently."

"仅在过去两个月里,每次会话的平均工具调用数就上升了约 30%。编码 agent 正在承担更复杂的工作:更频繁地读写文件、搜索代码、运行 shell 命令、浏览网页。"

"Since the start of 2026, the share of accepted AI lines still present after 60 minutes has risen from roughly 76% to 81%."

"自 2026 年初以来,被采纳的 AI 代码行在 60 分钟后仍然存在的比例,从约 76% 升到了 81%。"

"As models get more capable, use more context, and take on deeper tasks, cost becomes a bigger part of the product experience."

"随着模型变得更强、用上更多上下文、承担更深的任务,成本在产品体验中占的分量也越来越重。"

"To understand cost-intelligence tradeoffs, we looked at model economics from three perspectives: request cost, accepted-code efficiency, and the relationship between cost and benchmark performance."

"为了理解"成本与智能"之间的取舍,我们从三个角度审视了模型经济学:请求成本、被采纳代码的效率,以及成本与基准表现之间的关系。"

"Cost per agent request varies by nearly 9x across model families, showing that the same workflow can have very different cost profiles depending on the model behind it."

"每次 agent 请求的成本在不同模型家族之间相差近 9 倍——这说明同样一套工作流,背后换个模型,成本结构可以天差地别。"

"Cost per accepted line varies by roughly 7x across model families, compared with nearly 9x for cost per request, suggesting that higher-cost models partially make up the difference by producing more accepted code per request."

"按"每行被采纳代码"算,各模型家族之间的成本差约为 7 倍,而"每次请求"的差距是近 9 倍。这说明更贵的模型靠"每次请求产出更多被采纳的代码",把差距补回来了一部分。"

"This benchmark view plots model performance on Cursor's internal eval suite, CursorBench, against average task cost, showing where models sit on the cost-quality frontier."

"这张基准图把模型在 Cursor 内部评测集 CursorBench 上的表现,与平均任务成本放在一起,展示各个模型落在"成本-质量前沿"的什么位置。"

"AI is changing productivity across the board, but is concentrated at the top end of the user distribution."

"AI 在全面提升生产力,但提升高度集中在用户分布的顶端。"

"P99 power users are achieving significantly larger gains than everyone else, and the gap is growing much larger on an absolute basis as total AI usage grows."

"P99 的超级用户拿到的收益远大于其他所有人,而且随着 AI 总用量增长,这个差距在绝对量级上越拉越大。"

"AI usage is highly concentrated, with a small share of developers accounting for a large share of AI lines, spend, and token consumption."

"AI 用量高度集中:一小撮开发者就占掉了 AI 代码行、花费和 token 消耗中的一大块。"

"The Lorenz curves show this concentration, with Gini scores of 0.77, 0.75, and 0.72 across the three metrics, where higher scores on a 0-to-1 scale mean activity is more concentrated among fewer users."

"洛伦兹曲线(Lorenz curves)展示了这种集中,三个指标的基尼系数(Gini)分别为 0.77、0.75、0.72——在 0 到 1 的刻度上,分数越高,意味着活动越集中在更少的用户手里。"

"We see p90 developers pulling farther away from median developers in absolute lines added per week, with p99 users even farther out in the tail."

"我们看到 p90 开发者在"每周绝对新增行数"上越来越甩开中位数开发者,而 p99 用户更是远远甩在尾部之外。"

"P99 developers produce 46x more lines than the median active user and merge 15x more PRs than the median active PR author, while p90 users show large but much smaller gaps."

"P99 开发者产出的代码行是活跃用户中位数的 46 倍,合并的 PR 是活跃 PR 作者中位数的 15 倍;而 p90 用户虽然差距也大,但要小得多。"

"As models take on more complex work, they are using more context to understand the codebase, the user's intent, and the surrounding workflow before producing output. This shift has favorable cost implications because input tokens cost much less than output tokens, and cache-read tokens cost even less."

"随着模型承担更复杂的工作,它们在产出之前会用上更多上下文,去理解代码库、用户意图和周边的工作流。这一转变对成本是有利的,因为输入 token 比输出 token 便宜得多,而 cache-read token 更是便宜。"

"The shift toward context can help models produce better-calibrated code, consistent with the rising diff survival share we saw in developer productivity."

"向上下文倾斜,能帮模型产出"校准得更好"的代码——这与我们在开发者生产力部分看到的"diff 存活率上升"是一致的。"

"The ratio of input to output tokens is rising quickly, showing that models are consuming much more context for every token they produce. This shift suggests that models are doing more work up front before generating code."

"输入与输出 token 的比值正在快速上升,说明模型每产出一个 token,就要消耗多得多的上下文。这一转变意味着:模型在生成代码之前,做了更多的前期工作。"

"The same relative shift toward input tokens shows up in the token mix. Input now accounts for more than 90% of input-output token volume, making context the dominant part of non-cache model usage."

"同样的"向输入倾斜"也体现在 token 构成上。如今输入已占输入-输出 token 总量的 90% 以上,使上下文成为非缓存模型用量中的主导部分。"

"Input tokens dominate token consumption, but their effect on cost is moderated by their lower unit price. Even so, input tokens have become the majority of price-equivalent token costs, rising since the start of the year from roughly half of input/output token costs to nearly 70%."

"输入 token 主导了消耗量,但因为单价更低,它对成本的影响被压住了一部分。即便如此,按价格折算,输入 token 已成为成本的大头——从年初约占输入/输出 token 成本的一半,升到了接近 70%。"

"The context story becomes even larger once cache is included. Cache-read tokens dominate total token activity, showing how much agent work now depends on reusing prior context rather than reading everything from scratch. We continually improve our agent harness to best cache tokens across models and providers."

"一旦把缓存算进来,"上下文"这个故事会变得更大。cache-read token 主导了全部 token 活动,这说明 agent 的工作如今有多么依赖"复用既有上下文",而不是每次从零读起。我们持续打磨自己的 agent harness,以便在不同模型和供应商之间把 token 缓存做到最好。"

"AI coding tools started as a way to accelerate individual developers. We saw this impact in earlier sections of the report, with faster coding, larger PRs, deeper agent sessions, and more AI-generated code making it into commits."

"AI 编码工具最初只是给个体开发者提速的手段。我们在报告前面几节看到了这种影响:写代码更快、PR 更大、agent 会话更深、更多 AI 生成的代码进入了 commit。"

"Now AI software development is entering a new era, with AI becoming infrastructure for automating more of the software development lifecycle end to end."

"如今 AI 软件开发正进入一个新时代:AI 正成为一种基础设施,把软件开发生命周期中越来越多的环节端到端地自动化掉。"

"Since the start of the year, more than 5x as many agent-generated changes are reaching commits without a separate manual diff acceptance step, suggesting that developers are trusting agents to carry more work through the commit flow."

"自年初以来,无需单独的"人工逐行采纳"步骤就直接进入 commit 的 agent 改动,增加了 5 倍以上——这说明开发者越来越信任 agent,愿意让它把更多工作一路带过提交流程。"

"It is still early, but the first autonomy patterns are coming into focus. Adoption of Cursor Automations is growing quickly, with security review emerging as a strong automation use case."

"现在还很早,但最初的自治模式已经开始清晰起来。Cursor Automations 的采用增长很快,其中安全审查成了一个很强的自动化场景。"

"Even more recently, SDK runs show early demand for turning Cursor's agent infrastructure into a programmable platform customized to how each company builds software."

"更近一些,SDK runs 透露出一种早期需求:把 Cursor 的 agent 基础设施变成一个可编程平台,按每家公司构建软件的方式来定制。"

"This report provides a data-driven fixed point for understanding where agentic software development stands today, and where it appears to be headed next."

"这份报告提供了一个由数据锚定的坐标点,帮我们看清 agentic 软件开发今天走到了哪里,以及它接下来正朝哪个方向去。"