What happened
Vercel announced that Vercel Tracing has moved from beta to general availability on all plans. Users can search and inspect traces through the Vercel dashboard or CLI. The announcement says pricing is unchanged from the beta. The change is not only a status label: Vercel describes a consolidated Traces page, a shared query language for dashboard and CLI workflows, a Live mode, automatic tracing for certain new Next.js projects, and longer retention periods.
The dashboard’s Traces page presents every trace and span in one place, either for an individual project or for an entire team. Users can switch between a Spans list and a Traces list, filter the results, and search with Kibana Query Language (KQL). A Live mode streams new traces as they arrive. Vercel’s example query, span.name:db* AND span.duration > 500, selects database spans whose duration exceeds 500 milliseconds. In the beta, Vercel says, users could open a trace one request at a time from the Logs page; the new page is intended to make browsing and searching a broader set of trace data possible.
The CLI uses the same KQL as the dashboard, so a query can be carried between the two interfaces unchanged. vercel traces search --query 'span.duration > 2000' searches for spans slower than two seconds, while vercel traces search --status error --since 1h searches for errors from the last hour. Searches can cover every project in a team or be limited to one project with --project. Users can inspect an individual trace with vercel traces inspect <request-id> and manage a project’s sampling rules with vercel traces config.
Automatic tracing applies to new projects created on Vercel that use Next.js 16.5 or later and include at least one Node.js function. After the first successful build, Vercel turns tracing on, injects instrumentation at build time, and adds one sampling rule that traces 100% of requests in all environments. Vercel says those traces include Next.js framework spans without setup. Teams can opt out by removing all sampling rules in Settings → Tracing. This described default is for new projects; the announcement does not say that existing projects are automatically switched to it.
Retention is longer than during beta and now differs by plan: Hobby keeps traces for three days, compared with one hour before; Pro keeps them for 14 days, compared with one day; and Enterprise keeps them for 14 days, compared with three days. Vercel says beta traces followed runtime log retention. The announcement does not introduce a new price for the general-availability release.
What the archive adds
New Runtime’s earlier coverage on Python becoming generally available on Vercel (2026-10-21) and WebSocket support in Vercel Functions (2026-10-21) examined additions to Vercel’s runtime and platform surface. This is a new Vercel feature milestone, not a restatement of either launch: it adds trace search, retention details and a default for qualifying new Next.js projects.
Trend signal
This does not yet fit a pattern New Runtime tracks. The announcement concerns one hosting platform’s tracing workflow and defaults; it does not establish a broader shift in agent architecture, verification, or interoperability.
Why now
Vercel identifies this as the transition from beta to general availability. The accompanying changes address limitations it describes in the beta: traces were opened one request at a time from Logs, tracing required a sampling rule before data was collected, and retention followed runtime log retention. The announcement does not give a separate reason for the release timing.
Why it matters
For a team debugging a production request, the practical value is the ability to move from a slow or failed span in a search result to the surrounding trace, then use the same query in a terminal or dashboard. Searching across a team’s projects and streaming new traces can reduce the need to navigate logs one request at a time. Longer retention also gives teams more time to investigate intermittent failures, although the announcement does not quantify any reduction in debugging time.
The automatic default lowers setup effort for new Next.js projects on Vercel, and framework spans may help show where time is spent across a request. But tracing every request in every environment is a meaningful operational default, not a free guarantee of useful observability. Teams should check sampling settings and understand how their plan’s retention window fits their debugging needs. Existing projects should not assume they received the new default: Vercel specifically describes it for qualifying new projects.
This is platform observability rather than an AI-specific capability. It matters to AI builders insofar as their Vercel-hosted applications rely on Node.js functions and need to investigate request paths, including the services those applications call. The announcement does not claim tracing understands model behavior or provides purpose-built AI observability, so teams should treat it as request-level instrumentation, not a substitute for tracking model inputs, outputs, or evaluation results.
Limits and uncertainty
Vercel’s announcement describes the product and its stated defaults, but does not independently measure search latency, tracing overhead, or how much diagnostic work the new page saves. It does not provide pricing figures beyond saying pricing is unchanged from beta, nor does it quantify trace-volume limits or discuss the storage implications of retaining more data. The 100% sampling rule is described for new projects meeting the stated Next.js and Node.js conditions; the announcement does not say that existing projects change automatically. Teams should verify the settings and behavior in their own projects before relying on a particular retention window or instrumentation configuration.
Skill to build now
On a Vercel project, pick one slow or error-prone request path and make a reusable KQL query for it. Try the query in the Traces page and, if your project has CLI access, run the same filter with vercel traces search. Check whether the resulting trace includes the spans your team needs, confirm the project’s sampling rules and plan retention, and record a short runbook for inspecting a request by ID. For a new Next.js 16.5 project, explicitly review the 100% sampling default rather than assuming it matches your production or development policy.
Reader takeaway
Vercel Tracing is now generally available across plans, with search in both the dashboard and CLI, longer plan-specific retention, and automatic 100% tracing for qualifying new Next.js projects. The immediate action for Vercel teams is to test the shared query workflow and inspect sampling settings—especially on new projects—before treating the default as appropriate for every environment.