Benchmark · Measured
HTML to Markdown conversion
Webfold converts HTML to Markdown 3.9 times faster than Firecrawl's Rust html-extractor. That is 4.601 µs against 17.829 µs median. From a fresh process it is about 10.8 times faster. This times the conversion step only.
Webfold against Firecrawl's Rust html-extractor on 54 public pages.
What question does this benchmark ask?
How fast does each tool convert the same pages to Markdown? Do both pass the same quality checks?
How was it run?
- Reference: the 54 public fixtures from Firecrawl's Rust html-extractor at commit 0a2b7432.
- Five separate runs on one host (Intel Core i9-14900KF, CPU 8 affinity). Two warm-ups, seven samples, 50 conversions per hot sample.
- Hot timing excludes input loading. Fresh-process timing includes startup and input reading with a warm OS cache.
- Each figure is the median of five per-run aggregates.
What are the results?
| Measurement | Webfold | Rust html-extractor |
|---|---|---|
| Hot conversion, lower median of 54 fixture medians | 4.601 µs | 17.829 µs |
| Fresh-process conversion, same aggregation | 0.371781 ms | 4.021488 ms |
| Asserted fixtures passed | 52 of 52 | 52 of 52 |
| CPU-pinned re-run, 2026-09-30, hot (two runs) | 4.25 and 4.13 µs | 17.51 and 17.89 µs (4.1 and 4.3 times) |
| CPU-pinned re-run, 2026-09-30, fresh process (two runs) | 0.362 and 0.360 ms | 4.028 and 4.036 ms (11.1 and 11.2 times) |
Webfold is about 3.9 times faster on the hot path and about 10.8 times faster for a fresh process.
Both pass the same 211 expected-line and forbidden-phrase assertions on the 52 fixtures that have them.
A separate CPU-pinned re-run on 30 September, with the same fixtures, gave 4.1 and 4.3 times hot and 11.1 and 11.2 times from a fresh process, with Webfold at 1.000 on the asserted fixtures. The figures above are the five-run medians; the re-run agrees with them.
What does this not show?
Know the limits
- Conversion step only. Not the hosted Firecrawl service, networking or browser rendering.
- The fixtures are development fixtures that fixes were made against.
- The assertion score is not a measure of complete Markdown fidelity: a review found six malformed tables and one extra blank line that the assertions missed.
- Timings predate the current repair work, and CPU affinity was not exclusive. No confidence interval is claimed.
- An earlier run on 24 September, before the fixes, measured 5.0 times hot and 7.4 times fresh with Webfold at 0.955 on the assertions (10 forbidden strings leaked). The speed-up held; the quality gap closed.
Which other benchmarks are there?
- Document conversion Measured
- Main-content extraction quality Measured
- Web archive extraction Measured
- Footprint of a scrape or crawl Directional
- PII detection against Presidio Measured
- Localisation extraction Directional
- PDF text extraction with OCR Directional
- PDF extraction against other extractors Directional
- LaTeX conversion with BibTeX Directional
- A whole site to Markdown Directional
- Agent-browser compatibility (Lexbench) Measured
- Headless engine over CDP Directional
- Search-results parsing Directional
- Folder search Directional
Check it on your own pages.
Run the command-line tool on your content.