PDF to Markdown API in 2026

One price per document

A PDF read by the Extract API is one page read: $0.0008, the same as a web page. Firecrawl and the Apify actor count every page of the file, and pdfRest counts every call against a monthly plan. A file that gives no text comes back as empty, and a site that refuses the read comes back as blocked. Neither is charged.

Twenty files in one call

Send 1 to 20 URLs in urls, PDF, Word, Excel, EPUB, images and web pages mixed. Each comes back in data.results in the order sent, with its status, text, title and dates. The same URL twice is read once.

import os
import requests

reply = requests.post(
    "https://api.specstream.ai/v1/extract",
    headers={"Authorization": f"Bearer {os.environ['SPECSTREAM_API_KEY']}"},
    json={"urls": ["https://arxiv.org/pdf/1706.03762"], "format": "markdown"},
    timeout=60,
).json()

for doc in reply["data"]["results"]:
    if doc["status"] == "ok":
        print(doc["title"], doc["text"][:500])
print(reply["data"]["price_usd"])

Find the PDF and read it on one key

The Search API honors filetype:pdf, site: and after: in the query, and marks each PDF result with file_type. A fast search costs $1 per 1,000. Pass the PDF URLs to the Extract API and the agent has the text two calls later.

import os
import requests

H = {"Authorization": f"Bearer {os.environ['SPECSTREAM_API_KEY']}"}

found = requests.get(
    "https://api.specstream.ai/v1/web-search",
    params={"q": "EU AI Act guidelines filetype:pdf", "speed": "fast"},
    headers=H,
).json()
pdfs = [s["url"] for s in found["data"]["sources"] if s.get("file_type") == "pdf"]

read = requests.post(
    "https://api.specstream.ai/v1/extract", json={"urls": pdfs[:20]}, headers=H, timeout=60
).json()

Files behind a refusal

A site that refuses a plain read is tried again as a browser would ask for it, then in a real browser, then from the site's own feed, then from its newest public archive capture. read_from says which one gave the text, and captured_at dates an archive copy.

The top published score

The SpecStream Extract API scores 0.908 F1 on WCXB. The Search API scores 97.9 F-score on all 1,000 SimpleQA Verified questions and 64.0 item F1 on all 200 WideSearch tasks. Every score and its method is on the benchmarks page.

Rows from the whole web on the same key

Deep Search returns every match on the live web as structured rows at 5¢, 15¢ or $1 a row, with the source URL and read time on every value and the most a run can cost approved before it starts.

Firecrawl /parse

Takes an uploaded PDF, DOCX or XLSX file of up to 50 MB and returns Markdown, per-page content, layout blocks or JSON. PDFs cost 1 credit per page, and the Standard plan is $99 a month for 100,000 credits. SpecStream reads the file at its URL for $0.0008, and answers Firecrawl's own /v2/scrape call: see Firecrawl migration.

pdfRest

A REST tool that converts an uploaded PDF into a Markdown file. Plans run from $9.99 a month for 600 calls to $99.99 for 5,000, with overage at $0.115 or $0.06 a call; the free plan gives 100 calls a month with watermarked output. SpecStream bills $0.0008 a document with no plan.

Adobe PDF Services

Adobe's PDF Extract API has a PDF to Markdown job in its SDKs. The free tier gives 500 Document Transactions a month, and paid plans go through Adobe's sales team. SpecStream publishes one price, $0.80 per 1,000 documents, from the first call.

Apify PDF to Markdown actor

A community actor that turns PDF URLs or uploads into Markdown, text or JSON chunks, at $0.0011 a document plus $0.00012 a page, up to 500 pages a document. It reads PDFs only. SpecStream reads PDF, Word, Excel and EPUB files and web pages in one call at $0.0008 each.