murmr/Documentation
Try Playground

Rate Limits & Usage

murmr enforces rate limits per API key based on your plan. Limits reset monthly on your billing cycle date.

Per-Plan Limits

PlanCharacters/moConcurrent RequestsSaved VoicesAPI KeysWebSocket
Free50,000132No
Starter ($10)1,000,0003105No
Pro ($25)3,000,00052510No
Realtime ($49)3,000,00055010Yes
Scale ($99)10,000,0001010025Yes

Overage pricing

Paid plans can exceed their monthly character limit. Overage is billed at: Starter $15/1M, Pro $12/1M, Realtime $12/1M, Scale $10/1M. Free plans receive a 429 error when the limit is reached.

Character Counting

Characters are counted from the text (or input) field of your request. Whitespace and punctuation count. The voice_description field does not count against your quota.

Single request

A request with "Hello, world!" (13 characters) uses 13 characters from your quota.

Long-form generation

A 10,000-character article split into 3 chunks uses 10,000 characters total, not 3 requests. Characters are counted per chunk.

WebSocket

Each text buffer flush counts the characters in that chunk. A WebSocket session generating 5 flushes of ~100 characters each uses ~500 characters.

Concurrent Request Limits

Each plan has a maximum number of in-flight requests per API key. If you exceed the limit, the API returns 429 with details about the limit.

json
// 429 response for concurrent limit exceeded
{
  "error": {
    "type": "rate_limit_exceeded",
    "message": "Concurrent request limit exceeded",
    "concurrent_limit": 3,
    "concurrent_active": 3
  }
}

Wait for in-flight requests to complete before sending more, or upgrade your plan for higher concurrency.

WebSocket Limits

WebSocket connections have additional rate limits beyond character quotas.

LimitValueScope
Concurrent connections10Per API key
Generations per minute100Per API key
Auth timeout10 secondsPer connection
Max buffer size4,096 charsPer connection

If you exceed the generation rate limit, the server sends an error message on the WebSocket (code 4004) rather than closing the connection, so you can wait and retry.

Rate Limit Headers

Every HTTP API response includes headers showing your current usage:

HeaderDescriptionExample
X-RateLimit-LimitMonthly character limit for your plan1000000
X-RateLimit-RemainingCharacters remaining this period847250
X-RateLimit-ResetWhen the quota resets (ISO 8601)2026-03-01T00:00:00.000Z
typescript
// Check remaining quota from response headers
const response = await fetch("https://api.murmr.dev/v1/audio/speech", {
  method: "POST",
  headers: {
    "Authorization": "Bearer murmr_sk_live_...",
    "Content-Type": "application/json",
  },
  body: JSON.stringify({ input: "Hello", voice: "voice_abc123" }),
});

const remaining = response.headers.get("X-RateLimit-Remaining");
const limit = response.headers.get("X-RateLimit-Limit");
const reset = response.headers.get("X-RateLimit-Reset");

console.log(`${remaining}/${limit} characters remaining (resets ${reset})`);

Note

Rate limit headers are only included on HTTP API responses. WebSocket connections do not include these headers. Use the Dashboard usage page to monitor WebSocket usage.

Handling 429 Errors

When you exceed a rate limit, the API returns HTTP 429 Too Many Requests. Handle this with exponential backoff:

typescript
import { MurmrClient, MurmrError } from '@murmr/sdk';

const client = new MurmrClient({ apiKey: process.env.MURMR_API_KEY! });

async function generateWithRetry(input: string, voice: string, maxRetries = 3) {
  for (let attempt = 0; attempt <= maxRetries; attempt++) {
    try {
      return await client.speech.create({ input, voice });
    } catch (err) {
      if (err instanceof MurmrError && err.status === 429) {
        if (attempt === maxRetries) throw err;

        // Exponential backoff: 1s, 2s, 4s
        const delay = 1000 * Math.pow(2, attempt);
        console.log(`Rate limited. Retrying in ${delay}ms...`);
        await new Promise(r => setTimeout(r, delay));
        continue;
      }
      throw err; // Non-rate-limit error
    }
  }
}

For concurrent limit errors specifically, check the concurrent_active field in the error response to understand how many requests are in flight.

Monthly Reset

Character quotas reset on the first of each month (UTC). The exact reset timestamp is included in the X-RateLimit-Reset header on every API response.

Unused characters do not roll over to the next month.

See Also