Agent Data Platform

Not pretty, just extremely fast and useful. Machine-readable video generation pricing with provenance, freshness, and confidence on every row. REST + MCP. Free tier, no signup form.

What this is

One API answering one question well: what will this video generation job cost, across every provider that could run it? Currently 6 providers (fal.ai, Runway, Google Gemini API, Kling, Luma, Higgsfield), 20 models, ~29 pricing rows, normalized to USD per job. Every number carries:

Field	Meaning
`source_url`	Where the number came from. Provenance is not optional.
`verified_at`	When a human last confirmed it. Honest staleness — confidence degrades with age.
`data_confidence`	Source reliability set at entry: high (verified in-app / official docs), medium (trackers / support docs), low (conflicting or unverified sources).
`confidence`	The one to trust: MIN(data_confidence, freshness- and unit-derived confidence).

Changes are append-only in GET /v1/changes. Schema is date-versioned (meta.schema_version). Full spec: /openapi.json · agents: /llms.txt

Get a free key

100 calls/month, self-serve, returned once and stored only as a hash:

or: curl -X POST https://agent-data-platform.onrender.com/v1/keys/signup

Live demo

This calls the production POST /v1/estimate endpoint — the same one agents hit:

Duration s Resolution Count

REST in 30 seconds

curl -X POST https://agent-data-platform.onrender.com/v1/estimate \
  -H "X-API-Key: adp_..." -H "Content-Type: application/json" \
  -d '{"duration_seconds": 8, "resolution": "1080p"}'

Also: GET /v1/providers, GET /v1/pricing?provider=kling, GET /v1/changes?since=2026-07-01.

MCP (agent-native)

Streamable HTTP MCP server at /mcp with tools estimate_video_cost, compare_providers, get_pricing_changes. Same key, same metering as REST. Claude Desktop / Claude Code config:

{
  "mcpServers": {
    "video-costs": {
      "type": "http",
      "url": "https://agent-data-platform.onrender.com/mcp",
      "headers": { "X-API-Key": "adp_..." }
    }
  }
}

Pricing

Tier	Price	Limit
Free	$0	100 calls/month, 60 req/min
Paid (metered)	coming soon	contact 702executivellc@gmail.com for early access

Why trust this

Trust features are the product. Data is verified by hand against official pricing pages, in-app screenshots, and provider docs — and labeled honestly when it can't be (see data_confidence). Nothing is scraped blind. When a rate is login-gated or conflicting, the row says so instead of pretending. A weekly verification pass refreshes verified_at even when nothing changed, so staleness is visible, not hidden.

Run by 702 Executive LLC · 702executivellc@gmail.com