I’m currently evaluating your service for agent-based workflows and running into a fundamental limitation: the current rate limits make practical agent usage infeasible.
While the high token-per-second throughput is excellent and well-suited for low-latency generation, it is effectively offset by restrictive request rate limits. A single agent execution can easily exceed 60 requests per minute, which leads to throttling and interrupts the workflow.
In its current state, this creates a mismatch: the infrastructure supports fast responses, but the rate limits prevent sustained execution. This significantly reduces the value proposition and undermines the primary reason to adopt your platform.
Are there any options to increase the request rate limits—ideally within today—or to enable a configuration better suited for agent-style workloads?
Without an adjustment, I will need to consider moving this workload to another provider where the rate limits align with this use case.
@noltrix ,
Thank you for highlighting this issue. We understand that the current request rate limits are restricting agent-based workflows despite high token throughput. We’re reviewing options to better support sustained agent executions and will update you shortly.
Hi @noltrix ,
We understand that the current request rate limits are affecting agent-based workflows despite high token throughput. Could you please share the model, configuration, and workflow specifications you are using so we can better assess options? We’ll review and update you on potential solutions as soon as possible.
If you require higher we do have Enterprise tiers available. Would you like to have someone from our sales organization reach out to you about Enterprise tier commit plans?