One platform.
Every load test you'll ever run.
From a single recorded flow to elastic load across regions, with an AI copilot watching every percentile.
Author scenarios in a node graph.
Drag HTTP, gRPC, WebSocket, MQTT, and full browser steps onto the canvas. Branch on response data. Inject think-time. Record from a real session and let AI scaffold the assertions.
- HAR / cURL / Postman / OpenAPI import
- Per-step retries, timeouts, jitter
- Live preview with mock data
- Versioned in Git, PR-reviewable
From 100 VU to 5,000,000 VU. No infra to manage.
Geo-distributed across 18 regions with edge-warm runners. Spike, ramp, soak, or chaos shapes. Cold-starts under 2 seconds, billed by the VU-minute.
- 18 PoPs · AWS / GCP / Cloudflare
- Soak runs up to 72h with auto-checkpointing
- Custom load shapes via JS or YAML
- BYO runners on-prem (k6 + Playwright)
The bottleneck finds itself.
Trained on 14M historical runs. Detects regressions in real time, traces them to the offending commit, drafts the post-mortem, and suggests the smallest fix that holds the SLO.
- Real-time anomaly detection (P50/P95/P99)
- Run-over-run regression diffs
- Root-cause linked to PR + file
- Auto-drafted post-mortems
Dashboards that match your SLOs, not someone else's.
Build views from latency histograms, error budgets, throughput curves, and custom k6 metrics. Pin baselines. Block PRs that regress percentiles.
- Custom dashboards with shareable links
- Datadog / New Relic / Grafana export
- PagerDuty / Slack / GitHub Actions hooks
- Headless API for CI