概览
主要功能
- 逐步追踪输入、输出和 token 使用
- 创建数据集和自动化评估
- 内置、基于代码以及 LLM 作为评判者的评估
- 生产监控仪表板
- 人工反馈和注释收集
- 提示管理、版本管理和 playground
价格
- 模型
- Freemium
- 评分
- 4.8 / 5 (5)
使用场景
调试 LLM 应用链路追踪
检查详细的 LLM 链和代理的执行迹象 来识别故障、延迟瓶颈和意外输出
评估模型性能
在测试数据集上运行评估以测量质量、准确度和回归之前在生产发布更改
监控生产 LLM 应用
跟踪实时性能、使用和错误的部署的 LLM 应用以维护可靠性并快速诊断问题
优化提示工程
迭代提示并使用可观察性数据和评估指标比较版本来改进 LLM 应用结果
优点 & 缺点
优点
- 详细的嵌套链、代理和工具调用的追踪
- 集成的数据集和评估工作流来实现回归测试
- 紧密集成到 LangChain 和 LangGraph 中
- 生产监控成本、延迟以及反馈
- 框架中立的 SDK 可以超越 LangChain
缺点
- 最佳体验假设使用 LangChain 的生态系统
- LLM 作为评判者评估需要小心的设置和验证
- 商业用户量价格计划可以随之而生长
评测
5 个评分的平均值。
登录以留下评测。
Use it every day
Honestly didn't expect to like it this much. The automation is exactly what I needed, and the value for money is strong. I do wish the mobile experience lags, but I reach for it almost every day now and it just clicks.
Use it every day
Honestly didn't expect to like it this much. The dashboard is exactly what I needed, and support is responsive. I do wish the mobile experience lags, but I reach for it almost every day now and it just clicks.
Years in this space
I've evaluated a lot of these over the years. What stands out here is the integrations — handled better than most — and support is responsive. Worth the time if this is your use case.
Years in this space
I've evaluated a lot of these over the years. What stands out here is the onboarding — handled better than most — and the value for money is strong. Pricing gets steep at scale is my one real gripe. Worth the time if this is your use case.
Solid for our team
We rolled this out across the team last quarter and the value for money is strong. The onboarding fits neatly into how we already work, and the API removed a step we used to do by hand. The docs could be deeper, which is the main caveat, but it has held up under daily use.
问答
Does LangSmith Deployment include any free deployments?
Plus plans include 1 free small serverless deployment. If you spin up additional serverless or dedicated deployments, you’ll be charged on usage (resource time).
Asked by Priya Nair · Feb 6, 2026
What does uptime mean for LangSmith Deployment usage?
Uptime is the duration your deployment’s database is live and persisting state. Uptime will be tracked as soon as your deployment is live and ends when you shut it down. Dev agent deployments are typically short-lived (used during iteration, then deleted) – whereas Production agent deployments stay live and are updated via revisions (rather than being deleted).
Asked by Uma Krishnan · Feb 8, 2026
Why would I upgrade a base trace to an extended trace?
Base traces are short-lived (14-day retention) and ideal for quick debugging or ad-hoc analysis. They’re priced for volume and short-term utility. Extended traces are retained for 400 days. This is useful when traces include valuable feedback associated with them, such as from users, evaluators, or human labelers. This feedback makes them valuable for ongoing improvement and model tuning. LangSmith lets you choose the right retention for each trace, helping you balance cost and value.
Asked by Farah Rahimi · Feb 1, 2026
What is the difference between a base trace and an extended trace?
Base traces have a shorter retention period of 14 days. Extended traces have a longer retention period of 400 days. You can "upgrade" base traces to extended traces for an additional fee.
Asked by Hasan Demir · Jan 24, 2026
What is a trace? Can it contain multiple events?
A trace represents a single execution of your application—whether it’s an agent, evaluator, or playground session. It can include many individual steps, such as LLM calls and other tracked events. Here's an example of a single trace.
Asked by Anya Sokolova · Jan 18, 2026
提问
u5DE5\u4F5C\u5E38\u5F0F 的替代品
LangGraph Studio
u5DE5\u4F5C\u5E38\u5F0F
基于LangGraph的工作流可视化IDE
BrainSoup
u5DE5\u4F5C\u5E38\u5F0F
构建通过自然语言自动化任务和工作流的定制 AI 代理。
Letta AI
u5DE5\u4F5C\u5E38\u5F0F
开源平台用于构建具有长期记忆和先进推理的有状态人工智能代理
Snorkel Flow
u5DE5\u4F5C\u5E38\u5F0F
程序化数据标注和 AI 开发平台 - 使用标注函数更快地建立生产模型。
NetX
u5DE5\u4F5C\u5E38\u5F0F
模組化经济网络将区块链基础设施与人工智能能力紧密结合
Theoriq AI
u5DE5\u4F5C\u5E38\u5F0F
链上为构建和治理的多智能体 AI 系统提供的去中心化协议
Botpress
u5DE5\u4F5C\u5E38\u5F0F
端到端平台，用于构建、部署和管理 AI 代理和聊天机器人。
Gretel AI
u5DE5\u4F5C\u5E38\u5F0F
为 AI 准备的隐私安全合成数据平台
