translate5AI will be used in production by an external party. Their requests must always be processed first. Requests coming from translate5 (testing/internal) should run with lower priority and may wait when the server is busy with production traffic.

To enable this, the translate5AI server supports a priority parameter on each request. It controls the processing order: a lower number means higher priority, and requests without the parameter get the default value 0 (highest priority). When the server is under load, low-priority requests are queued or paused so high-priority requests are handled immediately.

Part 1 – translate5AI server:

Activate priority-based scheduling in the server configuration and restart the service. After this, the server sorts incoming requests by the priority value instead of plain first-come-first-served.

Part 2 – translate5:

Add the parameter priority with value 100 to every request body sent from translate5 to translate5AI:
 

$client->chat()->create([
    'model'    => 'mistral-small-4',
    'messages' => [...],
    'priority' => 100,
]); 

Production traffic sends no priority parameter, so it automatically gets 0 and is always served first. Add the parameter in one central place (HTTP client wrapper) so it applies to all translate5 requests.