Hacker News Firebase API vs Algolia: Which One to Use

Use the official Firebase API (https://hacker-news.firebaseio.com/v0/) for the front-page ranking (topstories), live lists and exact item lookups. Use the Algolia API (https://hn.algolia.com/api/v1/) for anything you need to search or filter by keyword, author, points or date. Firebase has no search and returns IDs you fetch one by one. Algolia returns full records in one call.

Both are free, need no key, and send Access-Control-Allow-Origin: *. I ran the same tasks through both on 2026-10-14. Here is how they differ.

Side by side

Firebase endpoints

Every path ends in .json. Without it, the server answered HTTP 301.

The item fields, per the official docs: id, deleted, type, by, time, text, dead, parent, poll, kids, url, score, title, parts and descendants. The docs say kids are "in ranked display order" and descendants is "the total comment count" of a story or poll. Only id is guaranteed; a field that doesn't apply is absent, not null.

Firebase also accepts its REST query parameters. topstories.json?orderBy="$key"&limitToFirst=3 returned only the first three IDs, and ?print=pretty indents the output.

Same task, both APIs

Fetching the current front page:

import time
from concurrent.futures import ThreadPoolExecutor

import requests

FB = "https://hacker-news.firebaseio.com/v0"
ALGOLIA = "https://hn.algolia.com/api/v1"
s = requests.Session()


def fb(path):
    r = s.get(f"{FB}/{path}.json", timeout=30)
    r.raise_for_status()
    return r.json()  # a missing item is HTTP 200 with the body `null`


# Official API: the ranked list, then one request per story.
t0 = time.time()
ids = fb("topstories")[:30]
with ThreadPoolExecutor(10) as ex:
    stories = [x for x in ex.map(lambda i: fb(f"item/{i}"), ids) if x]
print(f"Firebase: {len(stories)} stories in {time.time() - t0:.1f}s ({len(ids) + 1} requests)")
for rank, st in enumerate(stories[:3], 1):
    print(f"  #{rank} {st['score']} pts, {st.get('descendants', 0)} comments: {st['title']}")

# Algolia: one request for the stories on the front page now (not in rank order).
t0 = time.time()
hits = s.get(f"{ALGOLIA}/search", params={"tags": "front_page", "hitsPerPage": 50}, timeout=30).json()["hits"]
print(f"Algolia front_page: {len(hits)} stories in {time.time() - t0:.1f}s (1 request)")
print(f"  same IDs as Firebase top 30: {len({int(h['objectID']) for h in hits} & set(ids))}")

# Firebase list endpoints and how many IDs each returned today.
for name in ["topstories", "newstories", "beststories", "askstories", "showstories", "jobstories"]:
    print(f"{name:<12} {len(fb(name))} ids")

Output on 2026-10-14:

Firebase: 30 stories in 14.2s (31 requests)
  #1 524 pts, 211 comments: Iosevka is an open-source, sans-serif and slab-serif monospace font
  #2 25 pts, 3 comments: Submarine cable resilience
  #3 134 pts, 39 comments: Pg_vectorsearch: An index that is faster than pgvector's IVFFlat and HNSW
Algolia front_page: 30 stories in 3.0s (1 request)
  same IDs as Firebase top 30: 28
topstories   500 ids
newstories   500 ids
beststories  200 ids
askstories   22 ids
showstories  177 ids
jobstories   31 ids

Request times vary a lot from run to run. In an earlier test the same thread pool fetched 200 Firebase items in 4.4 seconds. The shape is what matters: Firebase is one request per item, and Algolia is one request per page of up to 1,000 items.

The two front pages overlapped on 28 of 30 IDs in this run and on 29 of 30 in an earlier one. The front page changes from minute to minute, and the two requests weren't simultaneous. If you need the exact ranking, topstories is the source: it's ordered, and front_page isn't.

Freshness

On 2026-10-14 the newest item in Algolia's /search_by_date had the same ID as Firebase's maxitem, 27 seconds after it was posted. Points and comment counts for the five top stories matched exactly between the two APIs. On that day Algolia was effectively live. Firebase is still the source of truth: Algolia indexes what Firebase publishes, so if the two ever disagree, trust Firebase.

Rate behaviour

• Firebase: the README says "There is currently no rate limit." That's not an invitation to hammer it. Use a small thread pool, as above, and cache items that no longer change (old stories and comments).
• Algolia: the API page says "We are limiting the number of API requests from a single IP to 10,000 per hour." That's about 2.7 requests per second, sustained. Because one request can return 1,000 hits, you rarely need that many.
• Neither publishes an error format for throttling. Back off and retry on 429 and 5xx.

Which one to use

Most real projects use both. Find IDs with Algolia, then read the authoritative item from Firebase when it matters.

Related guides for this API

• HN Algolia API reference.
• Get all comments for a Hacker News story, which compares all three ways of reading a thread.
• Search Hacker News by date range or points.
• The Hacker News Search API: free, no key.

Do it without code

The free Hacker News search tool runs Algolia searches live in your browser: keyword, type (story, comment, Show HN, Ask HN), relevance or date order, and a minimum score.

At scale

The Hacker News Scraper actor on Apify combines both APIs. Its list modes (top, new, best, ask, show and job) read the Firebase rankings, its search mode uses Algolia, and it can attach comment trees, author karma, and filters on score, comments, date and domain. It also parses the monthly "Who is hiring?" thread into job rows. It exports JSON, CSV or Excel, and it's free to start, then pay-as-you-go.

Disclosure: I build the datatooly tool and the Apify actor. Both APIs work fine without them.

Sources: Hacker News API on GitHub (official), HN Search API, and live responses from both APIs on 2026-10-14.