Source: ai-research/cartesia-sonic-product-page-2026.md, ai-research/cartesia-sonic-announcement-blog.md, ai-research/cartesia-sonic-3-5-docs.md, ai-research/cartesia-pricing-2026.md, ai-research/coval-tts-providers-2026-benchmark.md, ai-research/marktechpost-tts-models-2026-benchmark.md. 2026-10-30 field note: raw/Best_AI_Voice_Generator_-_HeyGen_vs_ElevenLabs_vs_Cartesia_vs_Inworld_vs_Fish.md (HeyGen official channel, vendor-made comparison).
Sonic is Cartesia’s real-time text-to-speech (TTS-only) model line — not a full-duplex conversational model like Moshi or OpenAI Realtime API, but a single pipeline stage typically paired with any LLM plus a separate STT stage, the same architectural role ElevenLabs occupies. Cartesia was founded by Albert Gu and Karan Goel, who (with Tri Dao) originated the S4 and Mamba state-space-model (SSM) architectures in academia before commercializing them — Sonic runs on Cartesia’s own SSM inference stack rather than a Transformer, which is the mechanical reason it markets itself specifically on latency.
Key Takeaways
- SSM architecture, not Transformer. Per Cartesia’s own launch claims (vs. a parameter-matched Transformer baseline): 20% lower validation perplexity, 2x lower WER, 1.5x lower time-to-first-audio, 2x faster inference, 4x higher throughput. SSM inference scales linearly with sequence length vs. Transformers’ quadratic scaling — the structural reason it stays fast under load.
- Version history moved fast: Sonic 1 (135ms model latency at launch) → Sonic 2 (40ms) → Sonic 3 (GA April 2026, ~90ms model latency / 190ms end-to-end per Cartesia) → Sonic 3.5 (May 2026, current recommended-stable release, sub-90ms marketed / ~82ms end-to-end time-to-first-audio per one independent benchmark) → a Sonic Turbo variant (~40ms). Briefly held #1 on the Artificial Analysis TTS leaderboard before being overtaken.
- Marketed latency and independently-measured latency disagree — this is a real, unresolved gap, not a rounding difference. ^[ambiguous] Cartesia’s own marketing claims sub-90–100ms “model latency” for Sonic 3.5. An independent production benchmark (Coval) measured Sonic 3 at P50 188ms, IQR 100ms, P75 269ms in practice — meaningfully higher and far more variable than the marketed figure, with a non-trivial share of requests exceeding the ~300ms threshold where conversational latency starts to feel laggy. Different sources also measure genuinely different things (model-latency vs. time-to-first-byte vs. time-to-first-audio vs. end-to-end vs. production P50), so cross-vendor latency numbers — including the comparisons in this article — are not strictly apples-to-apples.
- The ranking flips depending on which number you read. Using Cartesia’s own marketed figures, ElevenLabs Flash v2.5 (~75ms) beats Cartesia’s ~90ms claim. Using Coval’s independent production P50 data, Cartesia (188ms) beats ElevenLabs Turbo/Flash (264–288ms). Treat any single-number latency claim for any TTS vendor, including this one, with real skepticism until benchmarked in your own production conditions.
- Voice cloning: instant clone from a 10-second sample (per the product page; some secondary sources cite 3 seconds). Two tiers — Instant Voice Cloning (from the Pro plan) and Professional Voice Cloning (higher tiers, training fee).
- 42 languages natively, including 9 Indian languages. Sonic 3.5 added multilingual improvements (Hebrew, Japanese, Spanish, Hindi, German, Korean, French) and fixed long-session voice drift.
- Even a fast TTS leg doesn’t collapse a pipeline. Sonic makes the TTS stage faster, but in a pipelined stack (STT → LLM → TTS) the end-to-end budget per Coval’s analysis still runs 500–700ms — Sonic narrows one stage, it doesn’t eliminate the pipeline the way Moshi’s or OpenAI Realtime API’s single-model architecture does.
Pricing
| Tier | Price | Credits/mo | Notes |
|---|---|---|---|
| Free | $0 | 20,000 | — |
| Pro | 4 annual) | 100,000 | Adds commercial license + instant voice cloning |
| Startup | 37 annual) | 1,250,000 | Adds Professional voice cloning |
| Scale | 224 annual) | 8,000,000 | — |
| Enterprise | Custom | Custom | — |
Voice Agents (Cartesia’s “Line” product) bill separately: 0.014/min for telephony.
Implementation
Tool/Service: Cartesia Sonic (sonic-3.5 rolling-stable, dated snapshots like sonic-3.5-2026-05-04, or sonic-latest beta). REST/WebSocket API.
Setup: API key from Cartesia dashboard; third-party framework plugins exist, e.g. LiveKit’s livekit-agents[cartesia].
Cost: Free tier (20K credits/mo) up through 0.06/min + telephony surcharge.
Integration notes: TTS-only — pair with a separate STT model (e.g. whisper-cli for local/offline, or a cloud STT) and any LLM to build a full pipelined voice agent, the same shape as an ElevenLabs-based stack. $100M raised late 2025 (Kleiner Perkins, Index, Lightspeed, NVIDIA) — a well-capitalized vendor, not an early-stage risk.
Field note (2026-10-30): Sonic 3.6 in a HeyGen voice-cloning test
Source: AI Voice Generators Compared (HeyGen, 2026), a HeyGen-made video, so treat the ranking as vendor-biased.
- A newer version, Cartesia Sonic 3.6, was tested; this article’s body still describes Sonic 3.5. What changed in 3.6 is not in any source in the vault.
- At the video’s recording time the presenter put Sonic 3.6 sixth on the Artificial Analysis Controlled Voice Leaderboard, below HeyGen Voice Turbo, HeyGen Voice, Qwen Audio 3.1 TTS Plus and ElevenLabs v4 Turbo. This fits the “overtaken” note above, but the leaderboard named may not be the same TTS board cited there.
- In a cloned-voice test (same source recording, unseen script, no special prompting), the presenter scored Sonic 3.6 26.5/30, third of five: “clean and balanced,” identity close to the source, clear pronunciation and controlled pacing, with emotion “slightly more restrained” than HeyGen Voice or ElevenLabs v4 Turbo. The presenter’s recommendation: choose it for “a clean, balanced result without overly dramatic delivery.”
- The test covers clone quality for pre-rendered audio only. It does not address the latency questions above.
Related
- Moshi — Kyutai Labs’ Full-Duplex Speech Foundation Model — the open-weight, single-model full-duplex alternative to a Sonic-based pipeline
- OpenAI Realtime API — the other native-speech-to-speech option; unlike Sonic, doesn’t require pairing a separate STT/LLM stage
- ElevenLabs voice agents on Claude Code — the closest direct competitor: same pipelined-TTS role, closed/commercial, stronger on emotional/voice-cloning quality per independent comparisons
- Voice Agent Comparison — Moshi vs ElevenLabs vs OpenAI Realtime vs Cartesia Sonic — the full four-way comparison, including the marketed-vs-measured latency caveat in context
- HeyGen Voice — HeyGen’s video-adaptive voice-cloning model, which ranked above Sonic 3.6 in HeyGen’s own comparison
- whisper-cli (whisper.cpp) — a free/local STT stage that could pair with Sonic’s TTS stage in a self-hosted-leaning pipeline
Try It
- Before trusting any vendor’s marketed latency number (Cartesia’s or anyone else’s), test time-to-first-audio in your own production network conditions — the Coval benchmark shows marketed and measured latency can diverge by 2x or more.
- If building a pipelined voice agent, evaluate Sonic 3.5 against ElevenLabs Flash/Turbo on the specific latency percentile that matters for the use case (P50 vs P75 vs worst-case), not just the headline number.
- Use the Free tier (20,000 credits/mo) to benchmark voice quality and instant-cloning fidelity before committing to a paid tier.
Open Questions
- No source reconciled why Cartesia’s marketed model-latency and Coval’s measured production P50 diverge so widely (network conditions, request batching, geographic routing, and measurement methodology are all plausible factors, but none were confirmed in the sources gathered).
- Whether Sonic’s Voice Agents (“Line”) product is a viable alternative to the DIY STT+LLM+TTS pipeline for any use case this vault tracks (e.g. dental-client intake) is unexplored.