Firecrawl Alexandria 與 Treg：AI Agent 應選用哪種人員搜尋層？

Firecrawl Alexandria vs Treg: Which People Search Layer Should Your AI Agent Use?

Firecrawl Alexandria vs Treg comes down to how you want to pay and who picks the provider. Alexandria puts 430 providers, including 12 People sources such as Apollo, People Data Labs and OpenSanctions, behind your Firecrawl key, billed in plan credits. Treg puts 123 providers behind one token, routes each people call to a provider for you, and bills a prepaid balance at the provider's own rate, with misses at $0.

Try Apidog today

Firecrawl Alexandria vs Treg at a glance

Alexandria is a provider library you call by name. Treg is a router that picks a provider per call.

Catalog and coverage: who has which people data

Alexandria is wider; Treg is deeper on contact data.

The Alexandria library lists 430 providers and 2,461 capabilities across 20 categories. Its People category holds 12 providers: four commercial enrichment sources (Apollo, FullEnrich, People Data Labs, Data Legion), Particle for podcast speakers, and seven public-record sources, including OpenSanctions, the UK Sanctions List, the CMS NPI Registry and US Senate financial disclosures.

Those public records are the clearest gap. If your agent screens names against sanctions lists or checks an NPI number, Alexandria has a named tool; we didn't find one on Treg's people-search pages.

Treg's people-search page quotes 1B+ contacts across 123 providers, with Apollo, Hunter, Tomba and People Data Labs among them. Its routed capabilities cover the prospecting loop: treg.people.search, treg.people.email.find, treg.people.email.verify and treg.companies.enrich.

Our Firecrawl Alexandria explainer covers all 12 People providers in detail.

How each one routes a request

The biggest design difference is who chooses the provider.

Alexandria: you pick. Per the Alexandria docs, your agent discovers tools with POST /v2/search and sources: ["alexandria"], inspects one with find-tools, then runs it through POST /v2/scrape with an alexandria object:

curl -s https://api.firecrawl.dev/v2/scrape \
  -H "Authorization: Bearer $FIRECRAWL_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"alexandria": {"provider": "apollo", "capability": "people/search",
       "options": {"person_titles": ["VP of Sales"],
                   "q_organization_domains_list": ["example.com"], "per_page": 10}}}'

Results come back in alexandria[] with the payload under data. Some providers first return THIRD_PARTY_DATA_TERMS_REQUIRED with a requiresAction.url, and an org admin has to accept that provider's terms.

Treg: the router picks. You call a routed capability and Treg chooses the provider, cheapest first, and moves to the next one on a miss or a 429/5xx:

curl -X POST "https://treg.to/call/treg.people.email.find" \
  -H "X-Treg-[REDACTED CREDENTIAL] \
  -H "content-type: application/json" \
  -H "X-Treg-Route-Max-Cost: 0.05" \
  -d '{"full_name":"Jane Doe","domain":"example.com"}' -D -

The charge comes back in X-Treg-Cost-Micro (micro-USD). Direct tools such as apollo.people.search exist too, and don't fail over.

Want results from one known source? Alexandria's provider field pins it. Care more about the answer than its source? Treg's routing writes the fallback logic for you.

Pricing: plan credits vs provider rates

Alexandria spends the same credits as the rest of Firecrawl. Discovery is free. Execution costs the capability's listed credits. The Government and Research indexes are free up to a daily limit set by your plan. A firecrawl_feedback call within 20 minutes refunds 1 credit, capped at 10 per website and 100 per team per UTC day.

Plans from the Firecrawl pricing page, monthly billing:

Cost per credit is price divided by credits. Paid plans add credits in $5 increments: 1,000 on Hobby, 2,000 on Standard, 2,500 on Growth and 5,000 on Scale.

On November 3, the provider pages showed Apollo people/search at 0 credits and People Data Labs person/enrich at 60 credits. One PDL enrichment is $0.228 on Hobby and $0.0594 on Standard.

Treg has no subscription. It meters every call from a prepaid balance at the provider's rate with 0% markup. Misses on per-success providers bill $0.00. Your team's own provider keys are used first and never billed. New teams get $1.00 of credit once, and the cheapest email-find on the catalog starts at $0.0005.

Alexandria's credit price drops as your plan grows, which suits a team already spending Firecrawl credits. Treg's bill tracks answers, not plan size.

The same job on both: VP of Sales at 10 domains

The task: find the VP of Sales at 10 company domains and get a verified work email for each. We didn't run it; these are listed prices from November 3.

On Alexandria:

1. apollo / people/search with person_titles: ["VP of Sales"] and the 10 domains in q_organization_domains_list: listed at 0 credits.
2. A paid reveal or enrich step for each hit. Apollo's reveal capability price wasn't on the page we read, so have your agent run firecrawl find-tools --include-costs first. For scale, 10 PDL person/enrich calls at 60 credits are 600 credits: $2.28 on Hobby or $0.59 on Standard.

On Treg:

1. treg.people.search with company_domain and title for each domain.
2. treg.people.email.find for each person found, with X-Treg-Route-Verify: true to fold in a deliverability check.
3. Sum the X-Treg-Cost-Micro headers. Email-find starts at $0.0005, and misses cost $0.00. Treg publishes one receipt from September 23, 2026: 27 people, 21 emails, 20 verified, $0.58 total.

Run both on 10 of your own domains and compare cost per verified email.

Vendor benchmarks: read them as claims

Both vendors ran their own tests.

• Firecrawl: agents using Alexandria scored 21% higher on answer quality than built-in web tools across 845 tasks, with the same model and prompts and a blind AI judge.
• Treg, 100 leads (October 2026): Treg found 92 correct emails at 94% precision for $0.53. Clay, Apollo, Deepline and Hunter found 77 to 84, for $2.21 to about $3.35.
• Treg, LessieAI People Search Bench (119 tasks): Claude Code alone scored 43%; Claude Code with Treg scored 78.2%, at half the time per list and 38% fewer tokens.

Firecrawl's test measures answer quality; Treg's measure email accuracy and cost. Neither compares Alexandria to Treg. Our Clay alternatives breakdown goes through the 100-lead table in more depth.

Let your coding agent test both with Apidog CLI

Your agent's workflow depends on a response shape staying put and a price staying under a cap. A regression suite in Apidog CLI catches a renamed field or a pricier route before your prospect list does. You write one prompt; the agent runs the commands.

Set up the CLI and its agent skill once:

npm install -g apidog-cli@latest
apidog auth login --with-token <YOUR_APIDOG_TOKEN>
npx -y skills add https://apidog.com   # pick Claude Code, Codex or Cursor when prompted

Then give your agent this [REDACTED PROMPT]
Write an OpenAPI 3 file with two Firecrawl calls (POST /v2/search with
sources ["alexandria"], POST /v2/scrape with provider "apollo", capability
"people/search", per_page 5) and one Treg call (POST /call/treg.people.email.find).
Firecrawl sends [REDACTED CREDENTIAL]earer {{FIRECRAWL_API_KEY}}; Treg sends
X-Treg-[REDACTED CREDENTIAL] and X-Treg-Route-Max-Cost: 0.05. Import it into
Apidog project , create "Firecrawl" and "Treg" environments, and
write test cases that assert status 200, data.alexandria[0].data present with
no error, and X-Treg-Cost-Micro at or below 50000. Run each case against its
environment and summarize any failures.

The agent validates each payload and runs commands like these:

bash
apidog import --project --format openapi --file ./people-search-layers.json
apidog environment create Firecrawl --project --base-url https://api.firecrawl.dev
apidog environment create Treg --project --base-url https://treg.to
apidog cli-schema get test-case-create
apidog cli-schema validate test-case-create --file ./alexandria-apollo-case.json
apidog test-case create --project --file ./alexandria-apollo-case.json
apidog run --test-case --project -e \
--env-var "FIRECRAWL_API_KEY=$FIRECRAWL_API_KEY" -r cli,junit
apidog run --test-case --project -e \
--env-var "TREG_TOKEN=$TREG_TOKEN" -r cli,junit

The run stays cheap: Apollo search showed 0 credits and the Treg cap holds each call to $0.05. The same `apidog run ... -r junit` line drops into CI; the [Apidog CLI CI/CD guide](http://apidog.com/blog/apidog-cli-ci-cd-pipeline?utm_source=dev.to&utm_medium=wanda&utm_content=n8n-post-automation) has the pipeline files, and [Apidog CLI in Claude Code](http://apidog.com/blog/apidog-cli-in-claude-code?utm_source=dev.to&utm_medium=wanda&utm_content=n8n-post-automation) covers agent setup. Apidog's free plan covers up to 4 users.

## Which should you pick?

![Decision infographic: pick Firecrawl Alexandria if you already pay for Firecrawl credits, need public records such as sanctions, NPI and Congress data, want web search and provider data from one key, or choose the provider yourself; pick Treg if you want pay-per-answer with misses at $0, routing to the cheapest provider, open source code you can read or fork, or to plug in your own provider keys](https://assets.apidog.com/blog-next/2026/11/firecrawl-alexandria-vs-treg-pick-alexandria-or-treg.png)

**Pick Firecrawl Alexandria if:**

- You already use Firecrawl for search or scraping and want people data on the same credits.
- You need public records: sanctions lists, NPI, Senate disclosures, House roll calls.
- You want to name the provider on every call for reproducible results.

**Pick Treg if:**

- You want to pay per answer, with misses at $0 and no plan to size.
- You want routing and fallback handled for you, cheapest provider first.
- You already hold some provider keys and want them used first, free.
- You want the code open on GitHub.

If you're new to Firecrawl, start with our [Firecrawl CLI guide](http://apidog.com/blog/how-to-install-and-use-firecrawl-cli?utm_source=dev.to&utm_medium=wanda&utm_content=n8n-post-automation).

## FAQ

**What is the main difference between Firecrawl Alexandria and Treg?** Alexandria is a hosted catalog of 430 providers you call by name with Firecrawl credits. Treg is an open source router across 123 providers that picks the provider for you and bills at the provider's rate with 0% markup.

**Which is cheaper for finding work emails?** It depends on hit rate and plan. Treg bills misses at $0 and lists email-find from $0.0005. On Alexandria you pay the capability's credits at $0.0038 (Hobby) down to $0.000749 (Scale) per credit.

**Can I try both for free?** Yes. Firecrawl's Free plan gives 1,000 credits a month with no card, and Alexandria discovery is free. Treg gives new teams $1.00 of credit once.

**Does Treg cover sanctions or healthcare provider lookups?** We didn't find sanctions or NPI tools on Treg's people-search pages. Alexandria lists OpenSanctions, the UK Sanctions List and the NPI Registry in its People category.

**Are the benchmarks independent?** No. Firecrawl's 21% gain and Treg's 100-lead and 119-task results are each vendor's own test. Run both on your data first.