Rate Limit Calculator
Check if your usage fits within API rate limits for different LLM providers and tiers.
Your Expected Usage
| Model | Provider | Tier | RPM Limit | TPM Limit | RPD Limit | Status |
|---|---|---|---|---|---|---|
GPT-5.1 | OpenAI | Free | 3 3333% used | 40,000 250% used | 200 | Exceeded |
GPT-5.5 | OpenAI | Tier 1 | 500 20% used | 500,000 20% used | 10,000 | OK |
GPT-5.1 | OpenAI | Tier 1 | 500 20% used | 800,000 13% used | 10,000 | OK |
GPT-5 mini | OpenAI | Tier 1 | 3,500 3% used | 2,000,000 5% used | 10,000 | OK |
GPT-5.1 | OpenAI | Tier 2 | 5,000 2% used | 2,000,000 5% used | 100,000 | OK |
Claude Opus 4.8 | Anthropic | Tier 1 | 50 200% used | 30,000 333% used | 1,000 | Exceeded |
Claude Sonnet 4.6 | Anthropic | Tier 1 | 50 200% used | 50,000 200% used | 1,000 | Exceeded |
Claude Sonnet 4.6 | Anthropic | Tier 2 | 1,000 10% used | 100,000 100% used | 10,000 | OK |
Claude Haiku 4.5 | Anthropic | Tier 1 | 50 200% used | 50,000 200% used | 1,000 | Exceeded |
Gemini 3 Pro | Free | 5 2000% used | 250,000 40% used | 100 | Exceeded | |
Gemini 3 Flash | Free | 15 667% used | 1,000,000 10% used | 1,500 | Exceeded |
Understanding Rate Limits
- RPM: Requests Per Minute - Maximum number of API calls per minute
- TPM: Tokens Per Minute - Maximum number of tokens (input + output) per minute
- RPD: Requests Per Day - Maximum number of API calls per 24 hours
- Rate limits vary by provider tier - higher tiers have higher limits
- Implement exponential backoff and retry logic to handle rate limit errors
Understanding API rate limits
Every LLM provider caps how fast you can call its API, usually along three axes: requests per minute (RPM), tokens per minute (TPM), and requests per day (RPD). Hitting any one of them returns a 429 error, so a workload can be well within your budget yet still fail in production simply because it sends requests too quickly. Limits rise as you move up usage tiers and build a billing history.
This calculator checks your expected throughput against the published limits for each model and tier, flagging whether RPM or TPM would be the bottleneck. It's the fastest way to find out whether you need to request a tier upgrade, add client-side queuing and backoff, or spread load across keys before launch.
How to use this tool
- Enter your expected requests per minute and average tokens per request.
- Scan the table for any model or tier where your usage exceeds the limit — those rows are flagged.
- If you're over, plan for queuing, exponential backoff, or a higher usage tier.
Frequently asked questions
What's the difference between RPM and TPM?+
RPM caps how many requests you send per minute; TPM caps the total tokens across those requests. Long prompts hit the TPM limit first; many tiny requests hit the RPM limit first.
How do I raise my rate limits?+
Limits increase automatically as you move up usage tiers, which is driven by your cumulative spend and account age. Most providers also let you request a manual increase for production workloads.