Linkup pricing

About

Linkup (linkup.so) is a web search API built for AI agents and LLM applications. Rather than returning ten blue links, its endpoints return grounded, structured web data — raw search results, an LLM-composed sourced answer, or a schema-shaped structured object — that developers drop directly into agents, RAG pipelines, and other LLM-powered systems. The product surface is four endpoints: Search (synchronous web search), Fetch (URL content extraction), Research (asynchronous deep research), and Extract (a closed-beta structured crawl), plus a Tasks wrapper that batches any of them.

The company sells almost entirely to developers and AI teams, and positions itself as production-grade infrastructure “for enterprise AI teams who face real consequences when the model is wrong.” Public case studies cite leading AI labs and global financial institutions, including secure deployment for one of the world’s largest railway operators. SOC 2 Type II and Zero Data Retention are advertised on every plan at no additional cost.

Linkup’s pricing is deliberately simple: a single pay-as-you-go model with no seats, no plan tiers, and no annual commitment. Usage draws down a prepaid credit balance that automatically refills to $20 every month, so the “free tier” is a recurring monthly allowance rather than a one-time trial. Enterprise is the only sales-led motion, layering custom pricing, private/BYO-cloud deployment, personalized indexes, and SSO onto the same per-request engine.

Pricing summary : per-request web search API on a prepaid credit balance

Linkup uses a pure usage-based pricing model with no subscription tiers, billed against a prepaid credit balance — a structure adjacent to a freemium model thanks to its recurring monthly credit. The bill is built from three dimensions:

1. Per-request endpoint rate: Search costs $0.005 (standard/fast, raw results) to $0.05 (deep); Fetch costs $0.001–$0.005; Research costs $0.25–$2.50 by reasoning depth; the closed-beta Extract endpoint is variable (most tasks $2–$10, $10 minimum balance).
2. Output-type premium: Within Search, sourcedAnswer and structured outputs carry a small LLM-composition premium ($0.006 standard, $0.055 deep) over raw searchResults.
3. Prepaid balance: Every account holds a credit balance that auto-refills to $20 monthly (~4,000 standard searches); top-ups of $1,000+ earn bonus credits, and a 429 error is returned when the balance is exhausted.

What makes this different: there is no plan to choose and no seat to buy — the only lever is which endpoint and parameters you call, and the recurring $20 monthly refill makes the free allowance perpetual rather than a trial. Linkup also exposes agent-native x402 pay-per-request in USDC on Base, so an autonomous agent can transact with no account at all.

Pricing by product

Search endpoint (per request)

Fetch, Research and Extract endpoints (per request)

Account plans (no tiers)

Tasks and x402 (alternate access)

Sales motions across products: PLG / self-serve for all per-request endpoints and the $20 monthly credit; sales-led for Enterprise (custom pricing, private deployment).

Hidden costs : what an agent pipeline actually spends per month

The headline “$0.005 per search” and the free $20/month credit understate what a production agent pipeline spends once deep search, sourced answers, and research jobs enter the mix. Two real-world examples built from the published per-request rates:

RAG agent doing 50,000 sourced standard searches/month

Using sourcedAnswer instead of raw searchResults adds 20% ($0.006 vs $0.005) — a small per-call premium that compounds to $50/month at this volume.

Research-heavy agent: deep search + async research

Deep search and Research are where the bill concentrates: a single XL Research call ($2.50) costs as much as 500 standard searches, so endpoint and parameter selection — not search volume alone — drives spend.

Want to estimate your own Linkup bill? Use the Linkup pricing calculator to model your monthly cost based on search depth, output type, and Research/Fetch mix.

Pricing evolution : from search API to agent-native pay-per-request

Cadence

Tracked range: 2024-11 (founding) – 2026-07-14. Pre-March-2025 pricing surfaces were not preserved by Wayback; quarters not listed above were verified stable (0 price changes, 0 SKU additions).

Notable changes

• 2026-07-14 — Linkup added a closed-beta Extract endpoint (structured crawl, variable cost — most tasks $2–$10, $10 minimum account balance) and corrected the marketing pricing page’s Deep-search range from $0.05–$0.55 down to $0.05–$0.055 to match the billing docs; the actual per-request rate charged did not change (linkup.so/pricing, docs.linkup.so).
• 2026-06-04 — First UsagePricing capture: USD per-request pricing against the $20 auto-refilling monthly credit — Search $0.005–$0.05, Fetch $0.001–$0.005, Research $0.25–$2.50, x402 $0.01 flat (linkup.so/pricing, docs.linkup.so).
• 2026-05-05 — x402 agent-native pay-per-request (USDC on Base, no account) added; the pricing page was redesigned around endpoints (Fetch/Search/Research), retiring the Free / Pay-as-you-go / Custom plan cards and introducing the “Run 4,000 queries for free” framing.
• 2026-03-03 — $10M seed led by Gradient (angels from Mistral, Datadog, Deel, Dataiku); /fast sub-second search API launched. Pricing still on the EUR three-plan layout.
• 2025-03 — Earliest captured pricing (Wayback): EUR three-plan layout — Free (€5/mo credit ≈ 1,000 standard / 100 deep queries), Pay as you go (Standard €5/1,000, Deep €50/1,000), and Custom.
• 2024-11-28 — €3M pre-seed led by Seedcamp; Linkup launches as an ethical, licensed alternative to web scraping for LLMs, paying content partners per access (TechCrunch, 2024-11-28).

What’s unique : agent-native pay-per-request and a refilling free credit

1. The free tier is a recurring monthly allowance, not a trial. Linkup’s $20 credit refills to $20 every month, so a low-volume developer can run roughly 4,000 standard searches per month indefinitely without ever paying. That reframes “free” from acquisition bait into a standing usage budget.

2. x402 lets agents pay with no account. Linkup exposes pay-per-request over the x402 protocol in USDC on Base, billed at a flat $0.01/request. An autonomous agent can call the API and settle on-chain without a human ever creating an account — a genuinely agent-native billing path.

3. Output type is a priced dimension, not just a flag. Choosing sourcedAnswer or structured over raw searchResults carries an explicit LLM-composition premium ($0.006 vs $0.005; $0.055 vs $0.05), making response shape a deliberate cost lever rather than a free convenience.

Strengths & weaknesses

Billing UX : prepaid balance, top-ups, and consumption visibility

• Billing section in the Linkup app — add money to the prepaid balance and view current consumption.
• $20 monthly auto-refill — the credit balance is topped back up to $20 each month automatically; no action required.
• $1,000+ top-up bonus — adding $1,000 or more in a single top-up earns additional free credits.
• Charge-on-success only — no credit is deducted for failed requests (missing parameters, server errors, or no relevant results).
• 429 on exhaustion — when the balance runs out the API returns a 429 HTTP error until the account is topped up.
• Extract minimum balance — submitting a closed-beta Extract task requires a $10 minimum account balance; its variable cost is returned as creditsUsed in the task output.
• Custom-needs contact — [email protected] for custom billing arrangements; [email protected] for higher rate limits.

Strategic wins : pricing decisions that lower friction for AI builders

1. A refilling free credit beats a one-time trial for developer adoption

By topping the balance back to $20 every month, Linkup keeps low-volume developers on the platform without converting them to paid — and keeps them warm for the day their agent scales. This mirrors the perpetual-free-tier playbook more than a time-boxed trial.

2. Charging only on successful requests builds trust with agent builders

Agents retry, malform parameters, and hit empty results constantly. Not billing for failures removes a class of surprise charges that would otherwise erode confidence in a pure usage-based model.

3. Output-type premiums align price with cost

By pricing sourcedAnswer and structured above raw searchResults, Linkup recovers the LLM-composition cost exactly where it occurs instead of averaging it across all calls. This is the value-metric alignment that keeps a pure-usage model defensible as the underlying model costs shift.

Areas to improve : pricing clarity and large-buyer anchors

1. Make the closed-beta Extract cost predictable before general availability

Linkup closed its long-standing Deep-search discrepancy on 2026-07-14 — the marketing page now matches the billing docs at $0.05–$0.055 — but the newly added Extract endpoint reintroduces pricing opacity: cost is “variable,” most tasks land $2–$10, and the only guardrail is a $10 minimum balance with the charge revealed after the fact as creditsUsed. Before Extract exits closed beta, a published cost formula or per-complexity band would let teams budget a crawl the way they can already budget a Search call, in line with transparent enterprise pricing norms.

2. Publish a volume tier or anchor for Enterprise

Below the $1,000 top-up bonus there is no published volume curve, and Enterprise is fully gated. A simple “from $X/month” anchor or a posted volume discount band would help larger AI teams self-qualify before talking to sales, in line with transparent enterprise pricing norms.

3. Add spend caps and budget alerts to the prepaid balance

The prepaid model protects against runaway spend only by exhausting credit and returning a 429 — a blunt control. A configurable monthly spend cap and threshold alerts would let teams scale past the $20 refill without fear of a surprise top-up, a pattern covered in our spend-management guide.

Key takeaways

1. A refilling free credit can replace a trial. Linkup’s $20 monthly auto-refill turns the free tier into a standing budget, keeping low-volume developers engaged without a conversion deadline.
2. Make output shape a priced dimension. Charging a small premium for sourcedAnswer/structured over raw results aligns price with the LLM cost the response actually incurs.
3. Bill only on success in usage models. For APIs that agents call at scale, not charging for failures removes the single biggest source of bill-shock complaints.
4. Keep the pricing page and billing docs in lockstep. Linkup let a 10x Deep-search discrepancy ($0.55 vs $0.055) sit between its marketing page and billing docs before correcting it in July 2026 — a reminder that one premium-tier ambiguity can undercut the simplicity the rest of a pure-usage model works hard to achieve.
5. Expose an agent-native payment path. x402 pay-per-request in USDC lets autonomous agents transact without an account — a forward bet on machine customers.

UBP implications

1. Recurring free credits reframe the funnel. A monthly-refilling allowance blurs the line between freemium and pure usage, suggesting UBP vendors can use standing credits instead of trials to seed adoption.
2. Parameters, not plans, become the price menu. When endpoint and output-type selection drive the bill, the “pricing tiers” live in the API surface — pushing pricing strategy into product design.
3. Machine-payable APIs need machine-native billing. x402-style pay-per-request points to a future where usage-based pricing must settle with agents directly, not just human-held accounts.

Sources

• Linkup pricing page (accessed 2026-07-14)
• Linkup API pricing docs (accessed 2026-07-14)
• Linkup Enterprise (accessed 2026-07-14)
• Linkup rate limits docs (accessed 2026-07-14)
• Linkup x402 payment protocol docs (accessed 2026-07-14)

Bottom line

Linkup is one of the cleanest pure-usage stories in the corpus: no seats, no tiers, a recurring $20 monthly free credit, and per-request rates that range from $0.001 (Fetch) to $2.50 (XL Research). The agent-native x402 path and charge-on-success policy show a vendor designing for machine customers; the July 2026 Deep-search correction closed its last obvious clarity gap, leaving the variable-priced Extract closed beta as the one surface still worth watching.

Want to compare Linkup against other web search and data API pricing? Browse the pricing blueprint.

Pricing timeline : Major events on a vertical axis

Each milestone below corresponds to a public pricing change, product launch, or material adjustment. Major events use a filled marker; minor adjustments use a faded one.

Extract endpoint (closed beta) added; Deep-search advertised range corrected

Jul 2026

Linkup adds a closed-beta Extract endpoint (structured crawl, variable cost — most tasks $2–$10, $10 minimum balance) and corrects the marketing pricing page's Deep-search range from $0.05–$0.55 down to $0.05–$0.055, matching the billing docs. All other per-request rates unchanged.

USD per-request pricing with $20 auto-refilling monthly credit

Jun 2026

Current capture: prices presented in USD per request against a prepaid balance that refills to $20 monthly (~4,000 standard searches). Standard search $0.005–$0.006; deep search $0.05–$0.055; Fetch $0.001–$0.005; Research $0.25–$2.50 by reasoning depth; x402 at $0.01 flat. Startup credit grant is $5,000.

x402 agent-native pay-per-request; endpoint-centric pricing redesign

May 2026

Linkup adds x402 (Coinbase's open payment protocol) support, letting agents pay per request in USDC on Base with no account (2026-05-05). By the 2026-05-08 snapshot the pricing page was redesigned around endpoints — Fetch, Search, Research — replacing the Free / Pay as you go / Custom plan cards; 'Run 4,000 queries for free' framing introduced.

$10M seed led by Gradient; /fast sub-second search

Mar 2026

Linkup announces a $10M seed led by Gradient with angels from Mistral, Datadog, Deel and Dataiku, plus the /fast 'sub-second' web search API and NY/SF/Paris hiring (linkup.so/blog, 2026-03-03). The 2026-03 pricing snapshot still showed the EUR three-plan layout (Standard €5/1,000, Deep €50/1,000).

EUR three-plan pricing: Free / Pay as you go / Custom

Mar 2025

Earliest captured Wayback snapshot of linkup.so/pricing (2025-03-20) shows three named plans: Free (€5/month of credit ≈ 1,000 standard / 100 deep queries), Pay as you go (Standard search €5/1,000, Deep search €50/1,000), and Custom (contact sales). This EUR layout held stable through at least March 2026.

€3M pre-seed; content-licensing search API launches

Nov 2024

Linkup (Paris, founded 2024) raises a €3M pre-seed led by Seedcamp to build an ethical, licensed alternative to web scraping for LLMs, paying content partners per access (TechCrunch, 2024-11-28). Earliest pricing surface not preserved by Wayback before March 2025.