Is there a way to get page load time / performance metrics with Crawl4AI?

[hereyougo]

Hello,
Is there a way in Crawl4AI to get page loading timing metrics?

I see that DispatchResult contains start_time and end_time, but this seems to represent the overall crawl process, which may also include LLM processing and other steps.

Is there a any way to isolate or access page-level loading timings, or is seeing the total dispatch duration the only available option?

[hafezparast]

crawl4ai doesn't expose page-level load timing as a dedicated field, but there are two ways to get it with existing features:

1. Network request timing (capture_network_requests)

Enable network capture to get per-request timing, including the initial page load:

config = CrawlerRunConfig(
    capture_network_requests=True,
)

result = await crawler.arun(url, config=config)

# Find the main document request
for req in result.network_requests or []:
    if req.get("event_type") == "response" and req.get("url") == url:
        timing = req.get("request_timing")
        print(f"Page request timing: {timing}")
        break

The request_timing object contains Playwright's detailed timing breakdown (DNS, connect, TLS, TTFB, etc.).

2. Browser Performance API via js_code

You can extract the browser's native navigation timing directly:

config = CrawlerRunConfig(
    js_code="""
        window.__perf = JSON.stringify({
            dns: performance.timing.domainLookupEnd - performance.timing.domainLookupStart,
            tcp: performance.timing.connectEnd - performance.timing.connectStart,
            ttfb: performance.timing.responseStart - performance.timing.requestStart,
            download: performance.timing.responseEnd - performance.timing.responseStart,
            dom_interactive: performance.timing.domInteractive - performance.timing.navigationStart,
            dom_complete: performance.timing.domComplete - performance.timing.navigationStart,
            load_event: performance.timing.loadEventEnd - performance.timing.navigationStart,
        });
    """,
)

result = await crawler.arun(url, config=config)
# Access via result.js_execution_result

This gives you the standard Web Performance metrics in milliseconds — dom_interactive is when the page becomes usable, load_event is when everything (including images) finishes loading.

You're right that DispatchResult timing includes the entire pipeline (page load + extraction + screenshots). The approaches above isolate just the page loading portion.