Web search, distilled for your agents
One POST request returns clean, answer-ready web content. Mako feeds your LLM 2–3× fewer tokens than other search APIs — same speed, better answers. Moby dives deeper when you need the whole story.
Two models. Pick your depth.
Same request, one parameter. Mako skims the surface fast and lean; Moby dives to the bottom and brings everything back.
Surface · 0 mMakoThe fast one
Built for agents that search constantly.
Mako returns only the sentences that answer the question — a payload 2–3× smaller than other search APIs, so every search costs your LLM a fraction of the tokens. Same answer, smaller bill.
- Answers in ~1–2 s
- ~1k tokens per search
- Best default for agents & RAG
Deep water · −6000 mMobyThe thorough one
Built for research that needs the whole story.
Moby brings back the full readable content of every source, with a detailed answer that cites them. When one paragraph isn’t enough — deep research, due diligence, long-form synthesis.
- Full page content per source
- Detailed, source-cited answers
- Best for deep research
"model": "mako" → "moby"Search results your model can actually use
Most search APIs hand your agent a pile of links and noise. SERPdive hands it the content — clean, current, and already sized for an LLM.
Answer-ready content
Every result arrives as clean text — url, title, date, content — ready to drop straight into your model's context. No HTML, no boilerplate, no scraping on your side.
Live web, every time
Every search runs against the live web at the moment you ask — no cache, no stale index. What's true right now is what your agent gets.
An answer, not just sources
Turn on the answer and each search also returns a direct response to the question — concise on Mako, detailed and source-cited on Moby.
Speaks your users' language
Ask in French, get French sources and a French answer. The query's own language picks the right corner of the web — no locale settings to manage.
Fewer tokens, smaller bill
Search results become input tokens the moment your agent reads them. Mako's payloads run 2–3× leaner than other search APIs — the savings compound on every single search.
Integrated in minutes
One endpoint, two parameters: your question and a model. Plain HTTPS and JSON, no SDK required — the whole integration fits in a single function.
A search API bills you twice
Once per search call, then again when every result flows through your LLM as input tokens. SERPdive undercuts both bills: $4.50 vs $8 per 1,000 searches, and payloads 1.3× leaner.
$87.50 on search fees + $22.50 on LLM input tokens.
Pay-as-you-go rates: Tavily $0.008 per basic search, SERPdive $0.0045 per Mako search at the −25% launch rate (list $0.006). Token math assumes ~1,300 tokens per Tavily response vs ~1,000 for Mako, and Claude Sonnet 5 input at $3/M tokens. Median payloads — your queries will vary.
MakoOne request. Any language.
A single POST endpoint with two parameters: what to search and which model — Mako or Moby. Authentication is a bearer key. That's the whole integration.
- No SDK required — plain HTTPS + JSON
- Same response shape on both models
- Extracted page content, not raw SERP noise
curl -X POST https://api.serpdive.com/v1/search \ -H "Authorization: Bearer $SERPDIVE_KEY" \ -H "Content-Type: application/json" \ -d '{ "query": "what happened between Trump and FIFA", "model": "mako" }'
Judged blind, against the live web
Same question to both APIs, payloads anonymized, an independent LLM picks which content better answers it. No cherry-picking possible — the judge doesn't know who's who.
- 01
Questions no model can memorize
Live questions authored by a third-party frontier model — breaking news, market moves, live sport, product availability… the things an agent has to search because it cannot already know them.
- 02
Blind A/B duels
Both APIs get the same question; their raw payloads are anonymized and the A/B order alternates. The judge never knows which provider produced which content.
- 03
An independent judge
A frontier LLM at temperature 0, anchored to a reference answer from a third search engine. Stale facts presented as current, wrong-language sources and hallucinated pages lose the duel.
- 04
Public, end to end
The full 1,000-query benchmark ships with the judge prompt, every question and every per-duel verdict published — rerun it yourself and check.
Mako vsMako vsJuly 2026: 100 live questions per duel, GPT-5.4 judge at temperature 0 with an independent reference anchor. The public 1,000-query benchmark — prompts, questions and verdicts included — replaces these numbers when it lands.
Pay as you go, or commit and save. 25% off while we launch.
One rate, every feature. Start with 1,000 free credits — no card, no plan to pick.
Launch −25% ends
For new builders
- 1,000 API credits / month
- No credit card required
- Both models — Mako & Moby
- Email support
Pay only for what you use
- No monthly minimum, cancel anytime
- Production rate limits (1,000 req/min)
- Both models — Mako & Moby
- Failed searches are never billed
- Email support
4,000 credits / month · ≈ $0.0034 / credit
- Discounted rate — the more you commit, the less per credit
- Production rate limits (1,000 req/min)
- Both models — Mako & Moby
- Rollover unused credits within the month
- Priority email support
Frequently asked questions
The models, the pricing, the response format. Something we didn't cover? Ask us directly.
Ready to dive in?
Give your agents better answers for fewer tokens. 1,000 free credits, no card, and 25% off every search while the launch clock runs.