Parallel vs SerpApi vs Brave vs Bing in 2026

Run the same question through all four formats

Each API keeps its own request and its own JSON on SpecStream. This script sends one question in Parallel's, SerpApi's, Brave's and Bing's format with one key, and prints the result count, the seconds each call took and the first link.

import os
import time
import requests

KEY = os.environ["SPECSTREAM_API_KEY"]
BASE = "https://api.specstream.ai"
Q = "who maintains the requests library for Python"

calls = {
    "Parallel": lambda: requests.post(f"{BASE}/parallel/v1/search", headers={"x-api-key": KEY},
                                      json={"search_queries": [Q], "mode": "fast"}, timeout=30).json()["results"],
    "SerpApi": lambda: requests.get(f"{BASE}/serpapi/search.json",
                                    params={"engine": "google", "q": Q, "num": 10, "api_key": KEY}, timeout=30).json()["organic_results"],
    "Brave": lambda: requests.get(f"{BASE}/res/v1/web/search", headers={"X-Subscription-Token": KEY},
                                  params={"q": Q, "count": 10}, timeout=30).json()["web"]["results"],
    "Bing v7": lambda: requests.get(f"{BASE}/v7.0/search", headers={"Ocp-Apim-Subscription-Key": KEY},
                                    params={"q": Q, "count": 10}, timeout=30).json()["webPages"]["value"],
}

for name, call in calls.items():
    start = time.perf_counter()
    results = call()
    seconds = time.perf_counter() - start
    first = results[0]
    print(f"{name}: {len(results)} results in {seconds:.2f} s, first: {first.get('url') or first.get('link')}")

The four calls cost $0.004 together, from the $10 of credit a new account gets.

What 5,000 searches cost on each

Take an agent that runs 1,000 tasks a month with 5 searches each: 5,000 searches at list prices, before model tokens.

SpecStream: $5 on every path, page text included. $25 if every call asks Parallel's advanced mode.

Parallel: $5 on turbo or fast, and $25 on advanced, the mode a call gets when it names none.

SerpApi: $75 for the Developer plan of 5,000 searches a month.

Brave: $25, or $20 after the $5 of monthly credit.

Bing: the v7 API is gone. Grounding with Bing Search costs $70 at one tool call per search, and returns a model answer, not a list of results.

The most accurate web search API in the world

The SpecStream Search API scores a 97.9 F-score on all 1,000 official SimpleQA Verified questions, with 978 correct. On WideSearch it scores 64.0 item F1 and 43.4 row F1 on all 200 official tasks, and the Extract API scores 0.908 F1 on WCXB. The scores, the method and Parallel's published scores are at specstream.ai/benchmarks.

What each path reads and returns

Parallel: POST /parallel/v1/search reads search_queries, mode, advanced_settings.max_results up to 100 and source_policy, and returns results[] with url, title, publish_date and excerpts. The parallel-web SDK takes base_url="https://api.specstream.ai/parallel".

SerpApi: GET /serpapi/search.json?engine=google reads q, num from 1 to 100 and start, and returns organic_results[] with position, title, link, displayed_link and snippet, and serpapi_pagination.next on our host. In serpapi-python, set client.BASE_DOMAIN = "https://api.specstream.ai/serpapi".

Brave: GET /res/v1/web/search reads q, count up to 20 and offset, and returns web.results[] with title, url and description. The key stays in X-Subscription-Token.

Bing: GET /v7.0/search reads q, count up to 50 and offset, and returns webPages.value[] with name, url and snippet. Microsoft's SDKs build {endpoint}/bing/v7.0/search, and that path answers too, so an SDK needs only its endpoint set to https://api.specstream.ai.

Zero retention on any key

Set zero_retention when you create a key, or later. A search made with that key keeps no request text and no answer after the reply. The charge keeps the account, the key, the request ID, the price and the time.

curl -X POST https://api.specstream.ai/v1/account/keys \
  -H "Authorization: Bearer $SPECSTREAM_API_KEY" -H "Content-Type: application/json" \
  -d '{"name": "production", "scopes": ["account:read", "finds:spend"], "spend_cap_cents": 500000, "zero_retention": true}'

Search, pages and rows on the same key

The same key reads up to 20 pages a call with the Extract API at $0.80 per 1,000 pages, and runs Deep Search, which returns every match on the live web as structured rows at 5¢, 15¢ or $1 a row. The most a Deep Search run can cost is approved before it starts.

Tavily

Search for agents sold by credit: a basic search is 1 credit and an advanced search 2, at $0.008 a credit pay as you go. SpecStream answers Tavily's own POST /search at https://api.specstream.ai/tavily, in Tavily's JSON, at $0.001 a search.

Exa

Search by type: $4 per 1,000 requests on lite and $7 on auto and fast. SpecStream answers Exa's POST /search at https://api.specstream.ai/exa, at $0.001 a search of up to 10 results.

Serper

Google results as prepaid credits, from $1.00 per 1,000 on the 50k pack, with credits valid for 6 months. SpecStream answers Serper's POST /search at https://api.specstream.ai/serper, and a SpecStream top up never expires.

Firecrawl

Scrape, crawl and search by monthly credits: Hobby is $19 a month for 5,000 credits, and search costs 2 credits per 10 results. SpecStream answers Firecrawl v2 search and scrape at https://api.specstream.ai/firecrawl, at $0.001 a search and $0.0008 a page.