Voice Benchmark AI Agent Guidelines
Last-Updated: 2025-02-18

Preferred data endpoints (JSON):
- /api/rankings
- /api/providers/{id}/history
- /api/providers/{id}/history-full
- /leaderboard.json
- /providers.json
- /provider/{slug}.json

Update cadence & caching:
- Tests run hourly; cache responses for at least 60 seconds.
- Rate limit for third-party origins: ~120 requests/minute; please stay well below.

Data semantics:
- Latency fields ending in _ms are milliseconds.
- success is 1 for successful calls, 0 for failures.
- Provider slugs are canonical for URLs; ids are stable for API params.

Attribution:
- Cite “Voice Benchmark” with a link to https://voicebenchmark.ai when using rankings or stats.

Contact:
- For higher-volume access or questions: support+benchmark@dasha.ai