Your logs,
in context.
Logs next to your monitors and incidents. Send them from Vercel, AWS, OpenTelemetry or syslog, and search them for 30 days, for $0.25 per GB with no seats to count.
On every paid plan. Free while Logs is in alpha.
Trusted by teams that put uptime first.
Read customer stories
Wherever your code runs. One token per source.
Each source has its own token, retention and reception stats, so you see at a glance which service stopped sending. Lines are stored once the endpoint answers 202.
Connect a platform
- VercelLog drain, Pro and Enterprise
- AWS CloudWatchCloudFormation stack
- Cloudflare WorkersOTel export or Logpush
- HerokuHTTPS drain
- RenderLog stream over syslog
- Fly.iofly-log-shipper
- RailwayLocomotive sidecar
- SupabaseLog drain
- NetlifyLog drain
Or send them yourself
- OpenTelemetry
POST /v1/logs - HTTP JSON
POST / - Syslog over TLS
:6514
# One object, an array or NDJSON. Other fields become searchable attributes.
curl -X POST https://logs.hyperping.com/ \
-H "Authorization: Bearer $HYPERPING_LOGS_TOKEN" \
-H "Content-Type: application/json" \
-d '{"message":"Payment declined","level":"error","service":"checkout"}'Already shipping to Better Stack? The JSON endpoint reads Logtail payloads: change the URL and the token.
Your monitors say what broke. Your logs say why.
Logs live in the same project as your monitors, incidents and on-call, so the answer is already on the page when someone gets paged.
- api.acme.com · 502 · 3 regionscheckout · Timeout after 5000ms
The monitor fails, the logs explain
Every incident has a Logs tab, from 15 minutes before it started to 5 minutes after it ended. The warnings and errors your services logged are there, grouped by message.
Uptime monitoring - 16:14 · first timeout16:17 · outage opened
Before or after the outage?
The explorer draws your incidents over its histogram, so you see whether the errors started before the monitor failed or because of it.
Incident management - Open the war roomerrors · patterns · monitor · stream
A war room in one click
From an incident, build a dashboard of its window: errors of the linked service, patterns that appeared, the monitor and the live stream.
On-call scheduling
Find the line in a few keystrokes. Then everything around it.
Words, phrases, fields and attributes in one search bar, with suggestions from your own values. Press / from anywhere to start typing.
timeout dbBoth words, anywhere in the message"connection reset"The exact phrase-healthcheckLines without that word, phrase or fieldservice:(api|web)Any of the listed valueslevel:>=warnWarnings and worseattr.http.status_code:504Any attribute you send, nested ones includedresource.k8s.namespace:prodAn OpenTelemetry resource attributetrace:4bf92f35…Every line of a trace
- Live · 5,000 lines
Live tail that holds still
New lines stream in as they arrive, and wait while you scroll, hover or read one. A band counts what’s waiting.
- Lines around · Same pattern
Context in one click
The 20 lines before and after on the same host, every line with the same message shape, or every line of the same trace.
- ?q=…&from=…&line=…
Every view is a link
The search, the time range and the open line live in the URL. Paste it to a teammate and they see what you see.
- Errors · HTTP 5xx · Platform drains
Saved views
Built-in views for errors, warnings and 5xx responses, plus up to 50 of your own, shared by link.
- Filters · Columns
Filter without typing
Tick a level, service, host or source in the filter rail, and add any attribute as a column, such as the user ID or status code.
- CSV · NDJSON · Copy for AI
Take it with you
Export up to 10,000 lines, or copy a line and its context as Markdown, ready to paste into an assistant.
Dashboards that fill themselves. From the fields your logs already have.
Eleven templates read your sources first and tell you which ones fit. Pick one and its charts are live: no query to write, no field to map.
Logs overview
Volume by level and service, error rate, top patterns and sources. Works on any logs.
Errors and patterns
Errors by service against yesterday, top error patterns, and patterns new since last week.
HTTP access
Requests per second, 4xx and 5xx rates, latency percentiles, slowest and busiest routes.
Application (OpenTelemetry)
Requests by route, latency by service, error rate and exceptions by type.
Kubernetes / Docker
Logs by namespace, pod and container, OOMKilled and back-off restarts, chattiest pods.
Incident war room
Everything around one incident: its window, the linked service’s errors, the monitor and the stream.
Also: Linux host / syslog · Databases (Postgres, MySQL) · Jobs and workers · Vercel · Ingestion and cost.
Or write a chart in one line, with counts, rates, percentiles and a comparison with yesterday:
level:>=error | count() by service every 5m compare 1dYour AI reads them too. Through the Hyperping MCP server.
Claude, ChatGPT, Cursor and every MCP client read the same logs you do, scoped to your project, with the same search syntax.
- search_logs
Search from your editor
Claude, Cursor or ChatGPT search your logs with the explorer’s syntax, through the Hyperping MCP server.
- summarize_log_errors
Errors grouped by pattern
Recent errors folded into their message shapes, with counts, trend and first seen, so the assistant reads five problems instead of five thousand lines.
- get_outage_logs
Ask what broke
The assistant reads the lines around an incident and tells you where it started, next to the monitor’s own checks.
- run_log_chart
Charts on request
Error rate by service, p95 latency by route: the assistant charts your logs and answers with the numbers.
- Set it up with an AI agent
Wired by your coding agent
Each new source comes with a prompt for Claude Code, Cursor or Codex. The agent adds the logging setup to your project and checks that lines arrive.
- Copy for AI
One line, with its context
Copy any line with the lines around it as Markdown, ready to paste into the assistant of your choice.
Powering businesses of all sizes.From fintech to global agencies, teams know first.
Pay for the logs you send. Not for who reads them.
$0.25/GB ingested
On top of any paid plan, counted on the uncompressed lines we accept. Free while Logs is in alpha, until metered billing starts.
Estimate your billIncluded
- 30 days of search included
- Retention from 3 to 30 days, per source
- Dashboards and templates
- Logs on every incident
- MCP access for your AI assistants
Never billed
- A license per person who reads them
- Searches, live tail and exports
- Dashboards and charts
- Rejected lines
Questions about logs.
How do I send logs to Hyperping?
Create a source in Logs, copy its token, and point anything that speaks OpenTelemetry (OTLP over HTTP), HTTP JSON or syslog over TLS at logs.hyperping.com. Vercel, Heroku, Render, Fly.io, Railway, Cloudflare Workers, AWS CloudWatch, Supabase and Netlify send theirs through their own log drains. See Log sources.
How much does log management cost?
$0.25 per GB ingested, with 30 days of search included. Logs add no charge per seat, per search or per rejected line, and come with every paid plan. During the alpha, logs are free until metered billing starts. See pricing.
How long are logs kept?
You choose per source: 3, 7, 14 or 30 days. A search covers up to 7 days at a time, anywhere within the last 30.
How do logs show up on incidents?
Every incident page gets a Logs tab covering 15 minutes before the incident started to 5 minutes after it ended, with warnings and errors grouped by message. Switch to every line, or open the same window in the explorer. See Search and explore logs.
Is the format compatible with Better Stack (Logtail)?
Yes. The HTTP JSON endpoint reads Logtail payloads, so a shipper already pointed at Better Stack only needs a new URL and token. Common names for time, level, message, service and host are recognized, and every other field becomes a searchable attribute.
Do you store traces and metrics too?
Not as such. Hyperping stores logs, and keeps the trace and span IDs on each line so you can pull every line of a trace. Traces and metrics sent to the OpenTelemetry endpoint are accepted but not stored. Host metrics come from the server monitoring agent.
What does alpha mean for Logs?
Logs is open on every paid plan and runs in production, but features may still change. It isn’t part of the free trial yet, and accounts restricted to EU data residency can’t use it yet.
How is Hyperping different from Datadog or Better Stack logs?
Hyperping is not a full observability suite: there is no APM or error tracking. It puts your logs next to the monitors, incidents, status pages and on-call you already run, on one plan with no seat licenses. If you send terabytes a month or need tracing, a dedicated tool will fit better. Read the Datadog and Better Stack comparisons.