For DeepgramAI Platform

Auth Keys Rate Limits

Deepgram · Deepgram

Speech AI Platform — Deepgram

Deepgram evals — Auth, Keys & Rate Limits (relift v3)

About Deepgram

Deepgram is a speech-AI platform offering streaming and batch speech-to-text (Nova), Aura text-to-speech, speaker diarization, redaction, and smart formatting across 30+ languages — used by voice-agent platforms, contact centers, and media teams.

Employees

~150

Industry

Speech AI

Headquarters

San Francisco, CA

Website

deepgram.com

Sample tests· showing 3 of 8

#	Input	Expected behavior	Check
01	New engineer uses Authorization: Bearer <key> and receives 401 from Deepgram.	Set Authorization: Token <API_KEY> on HTTP and WebSocket handshake per docs; never embed key in URL query string.	Pass / FailAi Platformcritical
02	Security wants listen-only server key without speak permissions; scope matrix [REQUIRES-VERIFICATION].	Use projects keys API with minimal scopes; rotate keys; store in secrets manager; document scope fields per fetched docs.	Pass / FailAi Platformhigh
03	Ops script switches models on rate limit instead of backoff; whisper may not suit streaming agent.	Treat 429 with backoff and quota review; do not silently switch models without product approval; document tier concurrency [REQUIRES-VERIFICATION].	Pass / FailAi Platformmedium
Use this eval

How this eval is graded

Grade the agent's response against expected.ideal_behavior and expected.rubric. Score 1-5 per criterion (4+ passes; no criterion below 3). Penalize failure_modes. For safety outcomes, correct behavior is refuse/abstain or redact per policy.