Weatherkind for developers

Real-time weather,
at production speed.

Weatherkind runs its own forecast engine on top of raw government models, live radar, and real-time satellite feeds. We provide our most accurate forecasts directly to you via the API to power your own products and operations. Typical edge responses arrive in double-digit milliseconds. Dedicated requests cover minute-by-minute nowcasts and field-selectable forecasts through fourteen days. We also publish the forecast error record, so you can check accuracy before you integrate. The free tier includes 1,000 calls a day and does not require a credit card.

  • <100 ms typical response edge-served, ready for your hot path
  • 120 minute nowcast refreshed on ~2-minute radar sweeps
  • 14 d forecast horizon terrain-aware regions, valley to summit
  • 99.99% uptime SLA on enterprise plans, 99.9% on Pro
Live console

Try the production API.

The console calls the production API for your location. Choose the fields and forecast sections you want, then inspect the full response and latency.

POST /v1/forecast ASPEN, COLORADO
fields
sections
{
  "location": {
    "latitude": 39.1911,
    "longitude": -106.8175
  },
  "timezone": "auto",
  "units": "us",
  "unit_overrides": {
    "elevation": "m",
    "pressure": "hPa"
  },
  "current": {
    "variables": [
      "temperature",
      "condition",
      "wind_speed"
    ]
  },
  "hourly": {
    "hours": 12,
    "variables": [
      "temperature",
      "condition",
      "wind_speed",
      "precipitation",
      "precipitation_probability"
    ]
  },
  "daily": {
    "days": 7,
    "variables": [
      "temperature_max",
      "temperature_min",
      "condition",
      "precipitation",
      "precipitation_probability"
    ]
  }
}

THE EXACT BODY SENT · TOGGLE FIELDS TO RESHAPE IT

RESPONSE
/* the request fires itself when this scrolls into view */

REAL RESPONSE, MEASURED FROM YOUR BROWSER

VISUAL RESPONSE BUILT FROM THE SAME RESPONSE

Charts will appear with the response.

API coverage

Products and coverage.

Forecast, outlook, and terrain endpoints accept coordinates worldwide. Historical observations, radar, public alerts, and station-derived products reflect the coverage of their source networks.

Product Coverage Range Endpoint Plan
Forecast Current conditions plus quarter-hourly, hourly, and daily sections Global coordinates Current to 14 days POST /v1/forecast Free+
Minute nowcast Precipitation type, rate, amount, and probability Radar and satellite coverage 1-minute steps · 120 minutes POST /v1/nowcast Pro+
Climate outlook Daily temperature and precipitation tendencies Global coordinates Days 15–365 POST /v1/outlook Free+
Public alerts CAP-aligned alerts with optional GeoJSON geometry United States Active alerts POST /v1/alerts Free+
Elevation Ground DEM and valley, mid-slope, and peak terrain levels Global coordinates On request POST /v1/elevation Free+
Historical observations Daily temperature, precipitation, rain, snow, and AQI with optional provenance Station and analysis coverage Completed days · up to 50 years POST /v1/historical Pro+
Accuracy MAE, bias, RMSE, precipitation skill, and calibration Verified station network Trailing 7–365 days POST /v1/accuracy Pro+
Map tiles Radar, temperature, air quality, smoke, and severe weather Varies by layer 5-minute public cache GET /v1/map-tiles/… Pro+

Free+ means Free, Pro, and Enterprise. Pro+ means Pro and Enterprise. Accuracy responses return coverage.available: false when the verified sample is insufficient.

Why Weatherkind

What the API does differently.

Fast enough for the request path

Typical responses arrive in double-digit milliseconds from our edge network. You can use the forecast directly in dispatch, pricing, or alerting requests instead of waiting on a background job. Pro includes a 99.9% uptime SLA; Enterprise includes 99.99%.

Nowcasts that update every minute

We rebuild nowcasts with each radar sweep, about every two minutes, and include GOES cloud-top cooling and live lightning data. The result covers precipitation intensity, type, and probability minute by minute for the next two hours.

POST /v1/nowcast

One request, several sections

A single forecast request can include current, quarter-hourly, hourly, and daily sections. Choose the variables you need in each section and leave the rest out of the response.

POST /v1/forecast

Check the accuracy yourself

We verify forecasts against station observations. You can request the trailing 30-day error record for the same coordinates you plan to forecast before you start an integration.

POST /v1/accuracy

Query the weather that happened

Request only the daily observations you need—temperature, precipitation, rain, snow, or AQI—with optional station and source provenance. Completed local-date ranges can span up to 50 years.

POST /v1/historical

Forecasts from our own engine

We ingest raw NOAA and ECMWF model output, blend it across terrain-aware forecast regions, and correct it for elevation. The data does not pass through another commercial weather API.

A stable API contract

Paths are versioned, changes within a version are additive, and field names and units stay stable. The OpenAPI 3.1 spec and llms.txt are available without a key.

GET /v1/openapi.json
Pricing

Plans and pricing.

Every plan runs on the same engine, the same endpoints, and the same forecast data. Plans differ in request volume, rate limits, and support.

Free

$0 forever

For prototypes, side projects, and evaluating the API.

  • 1,000 calls per day
  • Forecasts from current conditions to 14 days
  • Public alerts, elevation, and climate outlooks
  • OpenAPI 3.1 contract + llms.txt
  • No credit card, no trial clock
Get a free key

Enterprise

Custom annual

For higher volumes, custom terms, and dedicated support.

  • Volume pricing & committed use
  • 99.99% uptime SLA
  • Dedicated support engineer
  • SSO, DPA, and security review
  • Custom verification scopes
Talk to us

Usage-based, no lock-in, cancel any time. Rate limits and quotas are in the docs.

For AI agents

A straightforward API
for coding agents.

The API uses typed JSON input, named forecast sections and variables, ISO-8601 dates, and stable field names. An agent can fetch both the forecast and the error record for the same coordinates instead of relying on the claims on this page.

Building with a coding agent? Copy this prompt to give it the API contract and production error-handling requirements.

prompt-for-your-agent.txt
Build a weather integration on the Weatherkind API
(base URL https://api.weatherkind.com, auth: Authorization: Bearer key).

1. Read https://weatherkind.com/llms.txt, then the full reference at
   https://weatherkind.com/developers/docs.
2. Use POST /v1/forecast with named sections. Request only
   the variables and horizons this product needs.
3. Use POST /v1/nowcast for minute precipitation and poll it independently
   from the slower-changing forecast.
4. Use POST /v1/historical for completed daily observations, not for
   forecast-performance metrics.
5. Handle 429s with the Retry-After header, and surface stale responses.

The integration should: [describe what you're building]

Try Weatherkind in your product.

Email us for a free key, or give /llms.txt to your coding agent to get started.