JTBD-scoped · comparable within a job
Leaderboards
Rankings are meaningful inside a job-to-be-done (or category). Raw overall scores are not comparable across categories — Stripe vs Anthropic is apples to oranges. Sunset providers excluded. As of 2026-10-19. These are editorial scores, not Arena Elo. Agents can filter minConfidence and region on the API. sort=price ranks only providers with a sourced price in the unit this job uses, and says how many were excluded. A chat $/1M rate is not used for embeddings, speech, or other non-token jobs. Throughput is still refused. Quote for this job: /api/quote · on the job page.
Coding agent / code generationFrontier general chat & reasoningSpeech-to-text / transcriptionProduction speech-to-textAgent RAG memory store
Showing JTBD “Production speech-to-text” · 6 providers · Depot Pick Deepgram
| # | Provider | Overall | Fit | $/perf | Latency | DX | As of |
|---|---|---|---|---|---|---|---|
| 1 | DeepgramDepot Pick | 8.5 | 8.8 | 8.8 | 8.6 | 8.6 | 2026-10-19 |
| 2 | Speechmatics | 8.2 | 9.0 | 7.2 | 8.8 | 7.8 | 2026-09-14 |
| 3 | OpenAI Media (Images/Audio) | 8.2 | 8.8 | 7.2 | 8.8 | 9.2 | 2026-09-14 |
| 4 | AssemblyAI | 8.1 | 8.6 | 7.2 | 8.4 | 8.8 | 2026-09-14 |
| 5 | Gladia | 7.9 | 8.2 | 7.8 | 8.0 | 8.2 | 2026-09-14 |
| 6 | Cartesia | 7.8 | 8.4 | 8.0 | 7.8 | 8.4 | 2026-09-14 |