Pricing · early access

Pick a passport class.

Same dataset, same API surface across all tiers. Higher tiers buy you more API calls, larger downloads, and direct lines for support. Prices in USD. Annual = 2 months free.

Two meters per tier: data API (city / POI / restaurant lookups, atomic Concierge sub-tools at /v1/agent/tool/*) and Concierge (full agent runs at /v1/agent/concierge). Concierge calls hit a cloud LLM and cost ~120× a data lookup, so they have their own monthly cap.

▸ VISA-FREE

Free

For evaluation + solo builders
$0 / month
5,000 data req / mo / data
50 Concierge req / mo / agent
  • All read endpoints
  • Concierge agent (capped)
  • Composite scoring access
  • Attribution required on public deployments
  • No SLA
  • No on-prem
Get a key →
▸ TOURIST

Starter

For shipping a single product
$49 / month
250,000 data req / mo / data
1,000 Concierge req / mo / agent
  • Everything in Free
  • No attribution requirement
  • Email support (48h)
  • API key rotation + scopes
  • No Parquet snapshots
  • No on-prem
Start at $49 →
▸ BUSINESS

Growth

For production AI agents
$249 / month
1,500,000 data req / mo / data
5,000 Concierge req / mo / agent
  • Everything in Starter
  • Parquet snapshot exports (/v1/snapshots)
  • Concierge SSE streaming endpoint
  • Slack support channel (24h)
  • 99.5% SLA
  • No on-prem
Get Growth →
▸ FIRST-CLASS

Scale

For high-volume agents + RAG
$499 / month
5,000,000 data req / mo / data
8,000 Concierge req / mo / agent
  • Everything in Growth
  • Graph traversals + semantic search unlocked
  • Concierge preflight (single-turn answers)
  • Priority request queue (front of line)
  • Dedicated rate-limit ceiling (1,500 req/min)
  • Weekly Parquet delta (not just full snapshots)
  • Slack support channel (8h)
  • No on-prem
Get Scale →
▸ DIPLOMATIC

Enterprise

For governments, OTAs, on-prem buyers
Custom · talk to sales
Unlimited data req / data
Unlimited Concierge req / agent
  • Everything in Scale
  • Enterprise-licensed data sources (otherwise gated)
  • On-prem deployment package
  • Weekly data-mirror inside your VPC
  • Local inference (Ollama, no cloud LLM calls)
  • 99% on-prem SLA / 99.9% cloud SLA
  • Dedicated solutions engineer
Talk to sales →
What breaks at each tier

Pick by failure mode, not feature list.

Most pricing pages tell you what you get. This one tells you what stops working as you outgrow each tier — so you choose based on what you'd actually hit.

  • Free →
    The 5,001st data request — or the 51st Concierge run — returns HTTP 429. Two meters, two ceilings. The Concierge cap (50/month ≈ 1-2/day) is the binding one for anyone testing the agent path; the data cap is generous enough for a demo. Upgrade is one click; no card on file required at signup.
  • Starter →
    Data meter caps at 250k/mo. A typical lookup-driven app (city autocomplete, POI details, restaurant search) doing ~8k req/day fits comfortably. A busy itinerary builder doing 5+ calls per session at 5k sessions/day does not.
  • Starter →
    Concierge meter caps at 1,000/mo (~33/day). Plenty for an internal tool or low-volume chatbot. If you're routing every customer message through the agent, you'll burn through it in the first week. Growth is the floor for production agent traffic.
  • Starter →
    Parquet snapshot endpoint returns 403. Bulk dataset exports for an analytics warehouse or a fine-tune corpus require Growth or above.
  • Growth →
    Concierge meter caps at 5,000/mo (~165/day). A real chatbot with 80-150 daily sessions averaging 1-2 agent runs each fits. A viral launch that 10×'s overnight does not — and the failure mode is loud (429 with a structured upgrade hint), not silent.
  • Growth →
    Graph traversals + semantic search return 403. The 6 graph tools (path_between, similar_destinations, country-to-country routing, themed clusters) and the embedding-based semantic-search route are Scale-and-above only.
  • Growth →
    You share rate-limit headroom with other tenants. Per-minute spikes from another customer can make your p99 wobble. Scale gives you a dedicated 1,500 req/min ceiling.
  • Scale →
    Concierge meter caps at 8,000/mo (~265/day). Most production agent products live here. If you're shipping 500+ Concierge runs/day across a real user base, Enterprise's unmetered Concierge — and on-prem local inference if you want predictable cost — is the next step.
  • Scale →
    Enterprise-licensed data sources are gated out. A subset of our data carries source-specific licenses that only ship to Enterprise contracts (audit trail per row available on request).
  • Scale →
    Your customer demands data residency. Tour boards, government tenders, EU enterprise — none of them want your queries leaving their VPC. On-prem ships only at Enterprise.
  • Enterprise →
    Nothing breaks. Custom usage tiers, dedicated engineer, NDA-fenced sandbox, local-inference Concierge (no cloud LLM calls). If you've never hit this tier, you probably don't need it.
Real workloads, real budgets

What it actually costs to run.

Concrete monthly cost for four real-shaped products. Numbers from real traffic profiles.

B2B DMC quoting tool · 80 dossiers/day

Marrakesh Inbound

Data: 80 dossiers × 10 lookups × 22 working days = 17,600 data req / mo
Concierge: 80 × 1 synthesis pass × 22 = 1,760 Concierge runs / mo
→ Growth ($249) — Concierge binds at 1,760/5,000, data is fine.
Hotel-POI overlay · 500 sessions/day, no agent in the loop

Booking-engine companion

Data: 500 sessions × 5.4 calls × 30 = 81,000 data req / mo
Concierge: 0 (atomic tools only, no full agent)
→ Starter ($49) — 3× data headroom, Concierge meter untouched.
Solo-dev side project · 30 chats/day on a personal site

Hackathon weekend

Data: 30 chats × 4 lookups × 30 = 3,600 data req / mo
Concierge: 30 × 0.4 agent runs × 30 = 360 Concierge runs / mo
→ Starter ($49) — Concierge binds (360 > Free's 50 cap).
Production AI chatbot · 1,000 sessions/day

Saaral Travel

Data: 1,000 × 3.2 atomic calls × 30 = 96,000 data req / mo
Concierge: 1,000 × 0.15 agent-routed × 30 = 4,500 Concierge runs / mo
→ Growth ($249) — Concierge close to 5,000 cap; jump to Scale if traffic doubles.
High-volume travel app · 18k sessions/day after viral launch

Saaral Travel (post-launch)

Data: 18,000 × 3.2 × 30 = 1,728,000 data req / mo
Concierge: 18,000 × 0.05 × 30 = 27,000 Concierge runs / mo
→ Enterprise — Concierge (27k) busts Scale's 8k cap. Custom contract with local-inference Concierge keeps unit cost flat.
National travel publisher · fact-check every draft

India Travel Weekly

Data: 40 articles × 12 direct fact-checks × 1 = 480 data req / mo
Concierge: 40 × 3 synthesis runs = 120 Concierge runs / mo
→ Starter ($49) — Concierge binds (120 > Free's 50). They take Growth anyway for the Parquet snapshots used in print.