Sonic key stats
- Latency (measured)
- 116 ms1
- Vapi streaming benchmark (50 trials per model) (checked 2026-06-10) Median of 50 sequential live streaming trials, June 2026; includes network RTT from the benchmark machine.
- docs.cartesia.ai/build-with-cartesia/tts-models/latest (checked 2026-06-10)
- cartesia.ai/pricing (checked 2026-06-10) 1 credit per character (docs.cartesia.ai/pricing); entry self-serve Pro plan is $5/mo for 100K credits, a $50 per 1M effective rate; larger plans drop to $37-39 per 1M. Same credit rate for every Sonic.
- cartesia.ai/blog/sonic (checked 2026-06-10)
Background
Sonic was Cartesia's debut voice model, released in May 2024 as the first text to speech engine built on the state space model architecture its founders pioneered in academia. At launch it generated lifelike speech with 135 ms model latency, the fastest in its class at the time, with instant voice cloning and speed and emotion controls. Three newer Sonic generations have since superseded it, but it remains the baseline that started Cartesia's run at real time voice.
Sources: cartesia.ai
At a glance
The first SSM voice model, with instant cloning and 15 languages. In our 50 trial streaming benchmark it returned first audio in a median of 116 ms, still among the fastest results on the Index.
Sources: docs.cartesia.ai
Frequently asked questions
- How is Sonic tested on the Humanness Index™?
- Listeners hear Sonic against another model in a blind head to head round, both voices reading the same customer support prompt from the same cloned source voice, and they pick whichever sounds more human. Its Humanness score derives purely from those votes.
- Is Sonic still Cartesia's current model?
- No. Three newer generations have superseded it, with Sonic 3.5 as the current flagship. Sonic remains in the arena as the baseline of the family.
Find the most human-sounding voice for your agent.
Compare the models in blind tests, read the methodology, or get in touch.
Build a TTS model? Add yours to the Index.
