Compare realtime APIs

Pick up to four APIs from any category and see them side by side.

Sarvam AI Saaras Realtime STT
Sarvam AI
Mistral Voxtral Mini Transcribe Realtime
Mistral AI
Voxist ASR (via OVHcloud / Scaleway marketplaces)
Voxist
CategorySpeech-to-textSpeech-to-textSpeech-to-text
StatusGAGAGA
Est. per minute$0.0059 - 0.0088$0.006$0.0063 - 0.016
How that was worked outINR 30-45/hr converted at about INR 85 per USD (third-party FX note, July 2026); currency risk applies.Flat announced rate; not confirmed on a live pricing table in this pass.Licence EUR 0.35-0.88/hr (~$0.006-0.016/min at ~1.1 USD/EUR) before GPU compute. Low confidence.
Pricing modelper-hourper-minuteper-hour
Free tierINR 100 credits for new users.Not stated.None stated.
Connects byWebSocketWebSocketWebSocket, REST
Audio inencoding and sample_rate parameters on the realtime channel (values not captured).AudioFormat(encoding='pcm_s16le', sample_rate=16000) in docs examples; mic example sends 480 ms chunks.WAV, mono only; examples use 16 kHz.
Audio outn/an/an/a
LanguagesIndian languages plus English (exact list not captured).13 languages (vendor).Not listed on the billing page (examples show en and French output).
Latency (vendor claim)Not captured.Vendor claim: configurable latency down to sub-200 ms; press reports ~1-2% error rate at 480 ms delay (secondary source).Not captured.
Key limits
  • See Credits and Rate Limits page (not captured).
  • rt_ client tokens expire after about 900 seconds; mint them just before connecting.
  • Python realtime needs SDK v2 (mistralai>=2, extra [realtime]); not available in v1 SDK.
  • WAV mono only.
  • Streaming has no diarization.
High-severity warnings
  • None
  • None
  • None
ComplianceNot re-verified.Not re-verified. Self-hosting keeps audio fully in your infrastructure.Runs in your own cloud project; certifications not checked.
Self-hostableNoYesYes
Last checked2026-10-102026-10-102026-10-10
Key numbers and features
$/hour$0.35$0.36-
Bills silence---
Free tierYes--
Free credit $---
Live speakers-NoNo
Turn detect---
Keyterms---
PII redact---
Partials-Yes-
Mixed langs---
8 kHz phone--No
Latency ms-200-
Languages-13-
Max session min---
Concurrency---
WebRTCNoNoNo
WebSocketYesYesYes
gRPCNoNoNo
HIPAA---
SOC 2---
EU data--Yes
Self-hostNoYesYes
Open weightsNoYesNo
High warnings000