Bulbul v4 Flash
Bulbul v4 Flash is Sarvam AI's latest text-to-speech model, released October 13, 2026 with the API ID bulbul:v4-flash.
Provider
Sarvam AI
Model family
Sarvam
Text-to-speech model
Cost tier
Bulbul
Status
Current
Release Oct 13, 2026
Technical specs
- Inputs
- text
- Outputs
- audio
- License
- Proprietary API
- API ID
bulbul:v4-flash- Voices
- 222
- Languages
- 11
- Price Inr Per 10k Chars
- 20
- Max Chars Per Rest Request
- 2,500
- Previous Price Inr Per 10k Chars
- 30
- Capabilities
- 200+ persona voices across Indian languages and English
- REST, HTTP streaming, and WebSocket APIs
- Under 250 ms time to first audio (vendor-reported)
- Same request shape as bulbul:v3
Why teams choose it
Complex reasoning
Useful for workflows that require structured thinking, multi-step logic, and deeper analysis than lightweight models provide.
Long-context analysis
Helps teams summarize, compare, and extract insights from long documents without losing important nuance.
Sarvam AI roadmap vigilance
Use published model pages—not stale marketing blurbs—for modalities, quotas, pricing, and policy; schedule revalidation tied to vendor release notes.
Cost-efficient routing
Useful as part of a routing stack where cheap models handle drafts and confirmations and this tier handles genuinely hard passages.
Tradeoffs to know
- bulbul:v3 stays the API default; pass model=bulbul:v4-flash explicitly.
- v3 speaker names such as shubh are rejected; use v4 persona IDs.
- 2,500 characters per REST request.
Benchmarks
- Time To First Audio Ms Max (vendor-reported)
- 250
Sarvam family lineup
Related models
Bulbul v4 Flash FAQ
What is Bulbul v4 Flash?
Bulbul v4 Flash is Sarvam AI's latest text-to-speech model, released October 13, 2026 with the API ID bulbul:v4-flash. It keeps the same request and response contract as bulbul:v3 on the /text-to-speech endpoint, adds a persona voice catalog of 200+ voices (222 speaker IDs), and Sarvam says time to first audio is under 250 ms. The docs list 11 languages (10 Indian languages plus English), REST, HTTP streaming, and WebSocket APIs, and 2,500 characters per REST request. Sarvam prices it at ₹20 per 10,000 characters, down from ₹30 for Bulbul v3. bulbul:v3 remains the API default when no model is passed.
When does Bulbul v4 Flash fit best?
Indian-language voice agents and IVR
What should teams watch out for with Bulbul v4 Flash?
bulbul:v3 stays the API default; pass model=bulbul:v4-flash explicitly.
Explore next
Models, tools, and comparisons that connect to this reference.