Capabilities
Explore what Saarika v2.5 supports, then use the API section for a request tailored to this model.
Languages
Model Information
- Provider
- Sarvam AI
- Product
- Speech-to-Text
- Release Date
- May 2025
- Data Policy
- Configurable
- Region
- India
- Input
- Audio
- Output
- Text
Playground
Try Saarika v2.5 by Sarvam AI. Requests use the selected workspace’s credits and API pricing.
Providers
Compare available endpoints, pricing, and context limits for Saarika v2.5.
| Provider | Context | Output | Input Price | Output Price | Endpoint |
|---|---|---|---|---|---|
| Sarvam AI | Not specified | Text | ₹0.425 / minute | — | sarvam/saarika-v2.5:sarvam |
Prices are catalog rates. Your workspace’s estimate and endpoint pricing determine the final charge.
Input and Output Limits
Formats and limits AudioLab enforces on each request.
| Limit | Value |
|---|---|
| Max bytes | Not specified |
| Max seconds | 30 |
| Formats | wav |
Provider Notes
Saaras v3 and Saarika v2.5 use Sarvam's native realtime API; saarika supports transcription, while Saaras also supports translate, verbatim, transliteration, and code-mixed modes. Saaras v3 Batch is a separate asynchronous product and is the only Sarvam slug here with speaker diarization. Its verified AudioLab input is WAV; MP3 is rejected before dispatch instead of failing later in Sarvam's batch job. AudioLab forces the realtime flush signal because session.finish uses manual provider drain semantics.
Performance
Status
Current catalog availability. Your workspace’s access is checked separately before a request runs.
A live availability check is not available right now.
Getting Started
Use your workspace API key and the exact model identifier below. Set the API URL to the endpoint shown in your workspace and keep credentials on your server.
curl "$AUDIOLAB_API_URL/v1/audio/transcriptions" \
-H "Authorization: Bearer $AUDIOLAB_API_KEY" \
-F 'model=sarvam/saarika-v2.5' \
-F 'file=@recording.wav'Request Parameters
Available parameters depend on the provider and request format. The endpoint reference explains accepted values and required fields.
| Parameter | Details |
|---|---|
model | sarvam/saarika-v2.5 — Required model identifier. |
mode (transcribe, translate, verbatim, translit, or codemix; Saaras) | View Accepted Values |
with_timestamps / with_diarization / num_speakers (batch) | View Accepted Values |
input_audio_codec (batch) | View Accepted Values |
high_vad_sensitivity / vad_signals / flush_signal (realtime) | View Accepted Values |
positive_speech_threshold / negative_speech_threshold | View Accepted Values |
speech frame, window, volume, interruption, padding, and ignored-frame controls | View Accepted Values |
Input and Output Limits
Formats and limits AudioLab enforces on each request.
| Limit | Value |
|---|---|
| Max bytes | Not specified |
| Max seconds | 30 |
| Formats | wav |
Similar Models
Explore more speech-to-text and compare their capabilities.
| Model | Context | Input Price | Output Price |
|---|---|---|---|
| Saaras v4Sarvam AI | Not specified | ₹0.425 / minute | — |
| saaras-v4-batchSarvam AI | Not specified | ₹0.425 / minute | — |
| Saaras v3Sarvam AI | Not specified | ₹0.425 / minute | — |
| Saaras v3 (Batch)Sarvam AI | Not specified | ₹0.425 / minute | — |
| MAI-Transcribe 2Microsoft Azure | Not specified | $0.0014 / minute | — |
| Gemini 3.5 TranscribeGoogle Gemini | Not specified | $0.0043 / minute | — |
