Rate Limits & Throughput

To maintain platform availability and prevent noisy neighbor saturation, Neosantara enforces a multi-layered rate limiter backed by Redis.

Three Rate Limiting Metrics

The Neosantara Gateway separates throughput controls into three independent metrics:

RPM (Requests Per Minute): Total HTTP call volume per minute enforced using a sliding window algorithm.
ITPM (Input Tokens Per Minute): Prompt token throughput enforced via a token bucket algorithm. Evaluated pre-flight before forwarding to the upstream model.
OTPM (Output Tokens Per Minute): Generated completion token volume. Settled post-flight after the model finishes producing output.

Throughput Limits by Tier

Account Tier	RPM (Requests/min)	ITPM (Input Tokens/min)	OTPM (Output Tokens/min)
Free	10 RPM	30,000 ITPM	8,000 OTPM
Basic	50 RPM	500,000 ITPM	80,000 OTPM
Standard	1,000 RPM	2,000,000 ITPM	320,000 OTPM
Pro	2,000 RPM	5,000,000 ITPM	800,000 OTPM
Enterprise	4,000 RPM	10,000,000 ITPM	1,600,000 OTPM
Coding Hemat	50 RPM	50,000 ITPM	10,000 OTPM
Coding Reguler	1,000 RPM	450,000 ITPM	90,000 OTPM
Coding Eksklusif	2,000 RPM	1,000,000 ITPM	200,000 OTPM

MCP Gateway Quotas & Limits by Tier

For IDE and agent integrations powered by the Model Context Protocol (MCP), Neosantara enforces dedicated request and summary token limits separate from gateway token throughput:

Account Tier	RPM Limit	Daily Requests	Max Summary Tokens
Free	20 RPM	500 req/day	900 tokens
Basic	40 RPM	1,500 req/day	1,200 tokens
Standard	60 RPM	3,000 req/day	1,600 tokens
Pro	120 RPM	8,000 req/day	2,200 tokens
Enterprise	240 RPM	20,000 req/day	3,200 tokens
Admin	Unlimited	Unlimited	6,400 tokens

Each account may provision up to 5 active MCP keys. Learn more in the MCP API Keys guide.

Promotional Model Limits (Free Tier)

For 100% discounted promotional models on the Free tier, a promotional limit of 5 RPM applies. To unlock the full Free tier limit (10 RPM), deposit at least IDR 50,000 via the billing top-up dashboard.

Handling Rate Limit Errors (HTTP 429)

When your application exceeds any of the limits, the gateway returns HTTP 429 Too Many Requests in standard JSON:

{
  "error": {
    "message": "You've reached your Input Tokens Per Minute (ITPM) limit for this tier. Please reduce your input length or wait a moment before retrying.",
    "type": "rate_limit_error",
    "code": "itpm_exceeded",
    "details": {
      "limit": 30000,
      "remaining": 0,
      "retry_after": 12
    }
  }
}

Client-Side Best Practices

Inspect the retry-after header to determine how many seconds to wait before dispatching the next request.
Implement exponential backoff with randomized jitter in your client network layer.
For bulk workloads that do not require real-time interactive responses, dispatch requests via Batch Processing which operates outside real-time RPM ceilings.

Guides

Operations & Security

Recursive Language Models (RLM)

Advanced AI Techniques

Three Rate Limiting Metrics

Throughput Limits by Tier

MCP Gateway Quotas & Limits by Tier

Promotional Model Limits (Free Tier)

Handling Rate Limit Errors (HTTP 429)

Client-Side Best Practices

​Three Rate Limiting Metrics

​Throughput Limits by Tier

​MCP Gateway Quotas & Limits by Tier

​Promotional Model Limits (Free Tier)

​Handling Rate Limit Errors (HTTP 429)

​Client-Side Best Practices

Three Rate Limiting Metrics

Throughput Limits by Tier

MCP Gateway Quotas & Limits by Tier

Promotional Model Limits (Free Tier)

Handling Rate Limit Errors (HTTP 429)

Client-Side Best Practices