Limits

Overview

These limits apply to POST /v3/routing/solve, /v3/routing/evaluate and /v3/routing/suggest. Most are checked before any travel times are fetched, so an oversized request fails fast and is reported by ?validate_only=true as well. A request over a count limit gets a 400 whose errors[] item has code limit_exceeded, the limit and the actual count your request has:

400 Bad Request

{
  "status": "BAD_REQUEST",
  "code": "invalid_request",
  "message": "the request has 10250 tasks (jobs + shipments); the maximum is 10000",
  "request_id": "req_5c1e8a3f9d02b746",
  "errors": [
    {
      "code": "limit_exceeded",
      "pointer": "/problem/jobs",
      "message": "the request has 10250 tasks (jobs + shipments); the maximum is 10000",
      "limit": 10000,
      "actual": 10250
    }
  ]
}

See Errors for the full error body.

Request

Limit	Value	When you exceed it
Body size	10 MiB, measured after decompression	`413`, code `payload_too_large`
Body compression	`Content-Encoding: gzip`, or none	`415`, code `unsupported_media_type`
Content type	`application/json`	`415`, code `unsupported_media_type`
Problems listed in one error response	100	The `message` says the list was cut; fix those and resend

Solve time

Limit	Value	When you exceed it
`options.time_limit_s` on solve	Default `1`. Maximum `300`. `0` is raised to 100 ms	`400`, item `limit_exceeded`, pointer `/options/time_limit_s`
`options.time_limit_s` on suggest	No default: the search runs to completion. Maximum `300`. `0` is raised to 100 ms	`400`, item `limit_exceeded`, pointer `/options/time_limit_s`
`options.max_suggestions` on suggest	Default `5`. `0` returns all. Maximum `100`	No error: a larger value is lowered to `100`

time_limit_s is whole seconds and counts search time only. Fetching travel times and waiting for a free solver are not deducted from it, so the request takes somewhat longer than the budget you set; set your HTTP client timeout accordingly. Evaluate takes no options because it does not search.

Problem size

Limit	Value	When you exceed it
Tasks: `jobs` plus `shipments`	10,000	`400`, item `limit_exceeded`, pointer `/problem/jobs` (`/problem/shipments` when there are no jobs)
`vehicles`	2,000	`400`, item `limit_exceeded`, pointer `/problem/vehicles`
Stops in total	65,535, counting every job, both stops of every shipment, a start and an end for every vehicle shift, and the reload visits each shift may make	`400`
Distinct locations	65,535	`400`
`quotas`	65,535	`400`, item `limit_exceeded`, pointer `/problem/quotas`
Reloads at one depot in one shift	8. The solver plans no more than that. List more than one depot in `shifts[].reloads` to allow more reloads in a shift	A `plan` with more is a `400`, item `invalid_value`, pointer `/plan/routes/{i}/stops/{k}`. This is checked when the plan is matched to the problem, so `?validate_only=true` does not report it

Capacity, skills and tags

Limit	Value	When you exceed it
Capacity dimensions	4 distinct names, counted across every `demand`, `delivery`, `capacity` and `initial_load` in the request	`400`, item `limit_exceeded`, pointer `/problem/vehicles`. The message lists the names found
Required skills	256 distinct combinations of skill `name` and `min_level` across all tasks. Skills that only vehicles have do not count	`400`, item `limit_exceeded`, pointer `/problem/jobs`
Task tags	256 distinct values across `jobs[].tags` and `shipments[].tags`. `serves_tags` and `preferred_tags` on the vehicle side do not count	`400`, item `limit_exceeded`, pointer `/problem/jobs`
`eligible_vehicles`	256 distinct vehicle sets across all tasks, after `allowed` and `excluded` are resolved. A set equal to the whole fleet does not count	`400`, item `limit_exceeded`, pointer `/problem/jobs`
Per-task `service_factors`	When any task sets it: vehicle shifts multiplied by task stops must not exceed 10,000,000	`400`, item `limit_exceeded`, pointer `/problem/jobs`

In a request with shipments and no jobs, the pointers shown as /problem/jobs in this table read /problem/shipments.

The four-dimension limit counts names, not vehicles. A typo such as wieght on one job counts as a dimension of its own. When you hit this limit unexpectedly, read the names in the error message.

Relations

Limit	Value	When you exceed it
Participants in one `synchronized` relation	8	`400`, item `limit_exceeded`, pointer `/problem/relations/{i}`
Tasks linked together across vehicles by `synchronized` and `time_lag` relations, and the `ordered` relations that join them	16 in one linked group	`400`
`ordered` relations that use `groups`	1 per request, with at least 2 groups	`400`

Values

Limit	Value	When you exceed it
Shift ids and depot ids	1 to 128 characters	`400`, item `invalid_value`, pointer to the `id`
`drop_fee`	At most 1,000,000,000	`400`, item `out_of_range`
`depots[].cost_per_visit`	At most 1,000,000,000	`400`, item `out_of_range`
Service time after `service_factor` is applied	At most 100,000,000 seconds	`400`, item `out_of_range`
`speed_factor`, `service_factor` and `service_factors[].factor`	From 0.001 to 5	`400`, item `out_of_range`
`limits.max_distance_m` and `limits.max_route_duration_s`	At most 2,147,483,647	`400`, item `out_of_range`

Prices

Amounts are in your currency. Prices have lower ceilings than drop_fee and depots[].cost_per_visit.

Field	Maximum	When you exceed it
`preferences[].violation_cost`	21,474.83 per entry, and 21,474.83 for all entries of one task added together	`400`, item `out_of_range`
`target.per_late_hour`	21,474.83	`400`, item `out_of_range`
`time_windows[].cost`	21,474.83	`400`, item `invalid_value`, pointer to the task’s `time_windows`
`time_windows[].per_late_hour`	1,288,490	`400`, item `out_of_range`
`objective.costs.per_late_hour`	1,288,490	`400`, item `out_of_range`
`objective.costs.per_travel_km`	21,474,836	`400`, item `out_of_range`
`objective.costs.per_route_hour`, `per_wait_hour` and `per_overtime_hour`	77,309,411	`400`, item `out_of_range`

For a rule that must always hold, use the hard field and not a high price: eligible_vehicles in place of a large violation_cost, and a time window with no lateness price in place of a large lateness rate. See Constraints.

Rate limits

Requests are also rate limited by the API gateway. A request over the rate limit is rejected with 429, and that response does not use the routing error body: see 429 Too Many Requests. The x-ratelimit-limit and x-ratelimit-remaining response headers report your rate limit and how much of it is left. When the service is at capacity it answers 503 with a Retry-After header. See Errors.

Errors

The error body and every code.

Performance

Choosing a time limit for your problem size.

Platform

Routing API

Overview

Request

Solve time

Problem size

Capacity, skills and tags

Relations

Values

Prices

Rate limits

Errors

Performance

​Overview

​Request

​Solve time

​Problem size

​Capacity, skills and tags

​Relations

​Values

​Prices

​Rate limits

Errors

Performance

Overview

Request

Solve time

Problem size

Capacity, skills and tags

Relations

Values

Prices

Rate limits