Batch Forecast API

Batch Forecast API

The new POST /v1/forecast/batch endpoint lets you submit up to 64 forecast requests in a single API call. Each request in the batch can target a different model, making it easy to compare models or forecast multiple series at once.

Request format

curl -X POST https://api.tsfm.ai/v1/forecast/batch \
  -H "Authorization: Bearer $TSFM_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "requests": [
      {
        "model": "chronos-bolt-small",
        "data": { "values": [1.2, 3.4, 5.6], "freq": "h" },
        "prediction_length": 24
      },
      {
        "model": "timesfm-2.0",
        "data": { "values": [7.8, 9.0, 1.2], "freq": "D" },
        "prediction_length": 7
      }
    ]
  }'

Response format

The response returns an array of results, one per request. Each result includes either the forecast output or an error, so partial failures do not block the entire batch.

{
  "results": [
    { "status": "ok", "forecast": { ... } },
    { "status": "error", "error": "Model not found: invalid-model" }
  ]
}

Limits

• Maximum 64 requests per batch call
• Each individual request follows the same validation rules as the single forecast endpoint
• Batch requests count toward your rate limit as individual requests

Batch forecasting is available on all plans.