Speech-to-text / transcription: best provider for agents (Depot Pick)

Scores as of 2026-10-19 (8 days old).

Depot Pick

Deepgram · 8.50/10 for this job (curated estimate, high confidence)

Blunt: openai-media was the Round-1 convenience pick before Deepgram entered the catalog. Deepgram wins production STT on $/min + streaming agent DX (Nova-3 / Flux). Keep openai-media only when you refuse a second vendor. Aligns with production-stt.

Runner-up

AssemblyAI · 8.11/10

Scores higher than the pick on: Developer & agent DX.

When not to use the pick

Already all-in on OpenAI and Whisper quality/latency is fine — openai-media; need richest audio-intelligence features (summaries/PII) — AssemblyAI; ultra-cheap batch Whisper on Groq LPU — groq.

Lock-in / migration

The pick has lower stored lock-in risk than runner-up AssemblyAI (4 vs 4.2). Self-host fallback listed for this job: OpenAI Whisper, whisper.cpp.

Sourced price (usd_per_minute)

Pick's lowest sourced SKU: Nova-3 Monolingual pre-recorded speech-to-text $0.0043 (usd_per_minute) — source, as of 2026-10-19

Cheapest matching SKU: Cloudflare Workers AI @cf/openai/whisper audio minute $0.0005 — source, as of 2026-10-03. Not the pick; not quality-adjusted.

Price history: first seen 2026-10-03 · last changed no change recorded · history JSON

Free option

• Cloudflare Workers AI speech-to-text (Whisper) (free) — source, as of 2026-10-19
• Gemini 3.5 Transcribe free tier (free tier unspecified) — source, as of 2026-10-26
• Groq Free Plan Whisper (free) — source, as of 2026-10-24

Failure modes to watch

• Runaway spend: Set a budget and a tool-call cap per task. Alert when spend jumps. Stop the run at the cap instead of summarizing a partial trace as success.
• Confirmation skip: Gate those tool names on an approval id recorded before the call. A missing approval is a failure even if the downstream tool returns success.
• Identity confusion: Bind every tool call to a user or tenant id. Refuse the call if that id changed mid-task. For a public bot, the operator owns the words the bot shows.
• Non-determinism: Run the task more than once and compare the end state, not a single transcript. Record the trial count next to any success rate you cite.
• Over-refusal: Log refusals separately from successes. Review a sample against the written policy. Do not treat a refusal as a pass, and do not weaken a safety check to raise the pass rate.
• Partial tool success: Require a per-item status and a failure count. Success for the task requires the failed count to be zero or explicitly handed to a retry queue.

Reported outcomes (submitted reports only)

This job: loading… JSON · reports never change scores · seed/test records excluded

Depot Pick (Deepgram) on this job: loading… JSON · reports never change scores · seed/test records excluded

Same-job, same-unit comparisons: Deepgram vs Cloudflare Workers AI

Cite: Agent Provider Depot, Depot Pick for speech-to-text: deepgram, scores as of 2026-10-19. Agents: GET /api/decide?jtbd=speech-to-text