WebThinker after the Bing Search API

One argument to the run script

run_web_thinker.py and run_web_thinker_report.py both declare --bing_endpoint with Bing's retired address as the default, and pass it to bing_web_search_async on every search. Before and after:

python scripts/run_web_thinker.py \
    --single_question "What is OpenAI Deep Research?" \
    --search_engine bing \
    --bing_subscription_key "YOUR_BING_KEY" \
    --api_base_url "YOUR_API_BASE_URL" \
    --model_name "QwQ-32B" \
    --aux_api_base_url "YOUR_AUX_API_BASE_URL" \
    --aux_model_name "Qwen2.5-32B-Instruct" \
    --tokenizer_path "PATH_TO_YOUR_TOKENIZER" \
    --aux_tokenizer_path "PATH_TO_YOUR_AUX_TOKENIZER"

python scripts/run_web_thinker.py \
    --single_question "What is OpenAI Deep Research?" \
    --search_engine bing \
    --bing_subscription_key "$SPECSTREAM_API_KEY" \
    --bing_endpoint "https://api.specstream.ai/v7.0/search" \
    --api_base_url "YOUR_API_BASE_URL" \
    --model_name "QwQ-32B" \
    --aux_api_base_url "YOUR_AUX_API_BASE_URL" \
    --aux_model_name "Qwen2.5-32B-Instruct" \
    --tokenizer_path "PATH_TO_YOUR_TOKENIZER" \
    --aux_tokenizer_path "PATH_TO_YOUR_AUX_TOKENIZER"

The key argument stays required: without it the script stops with "Bing search engine is selected, but --bing_subscription_key is not provided." The report mode takes the same two arguments.

The same request and the same JSON

WebThinker sends GET with q, mkt, setLang, textDecorations and textFormat, the key in Ocp-Apim-Subscription-Key. It reads name, url, snippet, siteName and datePublished from webPages.value, each with a default when it is absent. SpecStream answers in Bing's shape, 10 results a call at $0.005.

The same call from plain Python:

import os, requests

reply = requests.get(
    "https://api.specstream.ai/v7.0/search",
    headers={"Ocp-Apim-Subscription-Key": os.environ["SPECSTREAM_API_KEY"]},
    params={"q": "What is OpenAI Deep Research?", "count": 10},
).json()

for page in reply["webPages"]["value"]:
    print(page["name"], page["url"], page["snippet"])

Every parameter and field of the route is in the Bing Web Search v7 guide, and what Microsoft ended, and when, is on the Bing Search API retirement page.

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

A search of up to 10 results costs $0.005. 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. A busy call gets a 429 with Retry-After and charges nothing.

The most accurate web search API, in about one second

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.

Every Bing call runs the fast search. Measured on production on November 9, 2026, its median was 0.82 s at 1 search a second and 1.07 s at 20 a second, the 95th percentile stayed between 1.09 and 1.73 s, and it answered 718 of 720. There is no limit per key and no daily limit.

You decide what is kept, on each key

Create the key with zero_retention and a search made with it keeps no request text and no answer after the reply. spend_cap_cents is the most the key can spend, so a long WebThinker run stops at the cap you set.

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 returns every match for a request in plain words as structured rows, at 5¢, 15¢ or $1 a row, with a source URL on every value. The most a run can cost is approved before it starts.

WebThinker with Serper

WebThinker's own README says the Bing Search API "will be retired in August 2025" and adds --search_engine serper. Serper sells credits in packs, 50,000 for $50, valid for 6 months. SpecStream keeps --search_engine bing working, and its credit never expires.

Grounding with Bing Search

Microsoft's named replacement for the Bing Search APIs, at $14 per 1,000 transactions, inside Azure AI Agents. It returns an answer with citations, and developers get no access to the raw results, so WebThinker has no webPages.value to read. SpecStream returns Bing's webPages.value[] at $5 per 1,000.

Google Custom Search JSON API

$5 per 1,000 queries after 100 free a day, up to 10,000 a day, for existing customers only: Google closed it to new customers and ends it on January 1, 2027. It answers a different request with a different reply. SpecStream keeps WebThinker's Bing call and costs the same $5 per 1,000, with no daily limit.