GPT Researcher

Two variables, and the code stays as it is

GPT Researcher's searx retriever adds search to SEARX_URL, asks for format=json and reads results[] with url and content. Its serper retriever posts to SERPER_API_BASE plus /search with the key in X-API-KEY and reads organic[]. SpecStream answers both calls in those shapes.

# GPT Researcher .env: the two Google lines go
# RETRIEVER=google
# GOOGLE_API_KEY=...
# GOOGLE_CX_KEY=...

# the SearXNG retriever
RETRIEVER=searx
SEARX_URL=https://api.specstream.ai/searxng/YOUR_SPECSTREAM_KEY

# or the Serper retriever
RETRIEVER=serper
SERPER_API_BASE=https://api.specstream.ai/serper
SERPER_API_KEY=<YOUR_SPECSTREAM_KEY>

These are the calls each retriever sends:

# searx: it adds /search to SEARX_URL
curl -H "Accept: application/json" \
  "https://api.specstream.ai/searxng/$SPECSTREAM_API_KEY/search?q=state+of+solid+state+batteries+2026&format=json"

# serper: it adds /search to SERPER_API_BASE
curl -X POST https://api.specstream.ai/serper/search \
  -H "X-API-KEY: $SPECSTREAM_API_KEY" -H "Content-Type: application/json" \
  -d '{"q": "state of solid state batteries 2026", "num": 10}'

What you see when the Google key fails

The google retriever needs both GOOGLE_API_KEY and GOOGLE_CX_KEY, and raises Google API key not found without the first. When Google refuses the call, it prints Google search: unexpected response status: with the code, finds no items in Google's error reply, and returns an empty list. The research runs on, with no Google sources in the report.

Google closed the API to new customers, so a new project gets no key. With RETRIEVER=searx on SpecStream the same research run gets its sources back.

The reply your retriever already parses

The Serper route returns organic[] with title, link, snippet and position, as Serper does. The same call from Python:

import os, requests

reply = requests.post(
    "https://api.specstream.ai/serper/search",
    headers={"X-API-KEY": os.environ["SPECSTREAM_API_KEY"]},
    json={"q": "state of solid state batteries 2026", "num": 10},
).json()

for row in reply["organic"]:
    print(row["position"], row["title"], row["link"])

$5 per 1,000 searches, with no daily limit

A search of up to 10 results costs $0.005, and each started block of 10 more results adds $0.006. A research run sends many searches, so set spend_cap_cents on the key to cap what one run can spend. There is no monthly plan and no daily limit. Credit you buy never expires, and a new account starts with $10 of credit, which covers 2,000 searches.

The most accurate web search API in the world

On SimpleQA Verified the SpecStream Search API scores 97.9 F-score on all 1,000 official questions, 978 of them correct. On WideSearch it scores 64.0 item F1 and 43.4 row F1 on all 200 official tasks. The Extract API scores 0.908 F1 on WCXB. Every score and its method is on the benchmarks page.

About one second a search, measured under load

The SearXNG and Serper routes runs on the fast search. On November 9, 2026 production took 1, 5, 10 and 20 fast searches a second and answered 718 of 720. The median was 0.82 s at 1 search a second and 1.07 s at 20 a second, and the 95th percentile stayed between 1.09 and 1.73 s. One account can send up to 8,800 searches a minute; a call over that is answered with Retry-After and is not charged.

You decide what is kept, on each key

Give GPT Researcher a key with zero_retention and a search made with it 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. spend_cap_cents is the most the key can spend.

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 one key

The same key reads up to 20 pages or documents a call with the Extract API at $0.80 per 1,000 pages, and a page with no text is not charged. Deep Search takes a request in plain words and returns every match on the live web as structured rows, at 5¢, 15¢ or $1 a row, with a source URL on every value. The most a run can cost is quoted before it starts.

Google, until January 1, 2027

Existing customers can keep RETRIEVER=google until Google ends the API: $5 per 1,000 queries after 100 free a day, up to 10,000 a day. A new project gets no key. SpecStream costs the same $5 per 1,000 with no daily limit.

Vertex AI Search

Google's named alternative, for searching up to 50 domains, at $1.50 per 1,000 queries on Standard and $4.00 on Enterprise. A research run needs the whole web, and for that Google asks you to contact it. SpecStream searches the whole live web through searx or serper.

Tavily

GPT Researcher's default retriever, billed in credits at $0.008 each on pay as you go, with 1,000 free credits a month that reset on the first day of each month. Its address is fixed in GPT Researcher's code. With RETRIEVER=searx and SpecStream, a search costs $0.005.

Serper

Search sold in credit packs: $1.00 per 1,000 in the pack of 50,000, with credits valid for 6 months and 2,500 free queries to start. Keep RETRIEVER=serper and set SERPER_API_BASE to https://api.specstream.ai/serper: the same retriever reads the live web, on credit that never expires.