GamiWays
Cloud APICommercial

Cartesia Ink-2

WER 1.47% Pipecat (juin 2026). Modèle anglais ; le multilingue est annoncé, pas livré. TTS apparié : Sonic 3.6.

299ms
Latency (best case)
400ms
Latency (typical)
1.47%
WER (general audio)
$0.0072/min
Price per minute

Comparative Scores

Accuracy (WER)10/10
Streaming latency7/10
Multilingual7/10
Sovereignty2/10
Price accessibility7/10
Streaming quality8/10

Architecture

ArchitectureProprietary STT (Cartesia, San Francisco)
ParametersUndisclosed
Languages40+
Self-hostable No
Streaming Yes
WER clean audio 0%
GamiWays
Phase 1 MVP — Full-stack Cartesia pipeline (STT + TTS)

Excellent candidat pour un pipeline Storygami/Edugami full-stack Cartesia : Ink-2 STT (#1 WER) + Sonic 3.5 TTS (#4 ELO, 40ms TTFA). Combinaison latence+qualité optimale pour agents vocaux temps réel. Pas de souveraineté (cloud US) — évaluer pour Phase 1 MVP, pas pour déploiement souverain.

Analysis

Cartesia Ink-2 achieves the best WER on the Pipecat STT Benchmark (June 2026) at 1.47%, ahead of Deepgram (1.71%) and AssemblyAI (1.74%). Priced at $0.43/hr with 299ms TTFS. As of 30 September 2026 Cartesia still describes this voice-agent model as English-only, with multilingual support announced and not shipped. The paired TTS on this portal is now Sonic 3.6, which does speak French. Ink-2 is therefore not a French STT for Où est Ava ?.

Strengths

  • WER 1.47% — #1 Pipecat STT Benchmark juin 2026
  • Pipeline full-stack avec Sonic 3.5 TTS (ELO 1203)
  • $0.43/hr — prix compétitif
  • 40+ langues
  • Intégration native Cartesia (STT + TTS)

Weaknesses

  • Cloud only — pas de souveraineté
  • Pas de diarisation
  • Pas de Voice Agent API intégrée
  • Pas de score ELO sur Artificial Analysis

STT Capabilities

Streaming Yes

Streaming WebSocket. 299 ms TTFS (Pipecat, juin 2026). La page Ink décrit un modèle anglais, multilingue à venir. Le TTS apparié courant est Sonic 3.6, pas une voie française complète.

Diarization No
Custom Vocabulary No
Word Timestamps Yes
Auto Punctuation Yes
Multilingual Yes

40+ languages

Pricing

Price / minute
from $0.0072
depending on plan
Price / hour
from $0.432
depending on plan
Free tier
10,000 credits (no commercial use)

$0.43/hr ($0.0072/min) Ink-2 streaming (Pipecat benchmark juin 2026). Cartesia Pro: $5/month + 100K credits.

Hybrid (PAYG + Subscription)
Official pricing page
PlanSubscription/moIncludedOverage/minMax streamsTop-up
Free

Upgrade required for more streams

Free20K credits$0.00722—
Pro

Extra credits at same Pro rate

$5/mo100K credits (~7 min audio)$0.00723Yes
StartupRecommended

Extra credits at Startup rate

$49/mo1.25M credits (~83 min audio)$0.00655Yes
Scale

Extra credits at Scale rate

$299/mo8M credits (~533 min audio)$0.005515Yes
Enterprise

Custom concurrent limits

FreeCustom—∞Yes
Parallel stream limits: The number of simultaneous streams is limited per plan. For events with many concurrent users, verify the required plan or contact the provider for an Enterprise agreement.
Top-up available: Some plans allow purchasing additional minutes (top-up) without changing plans. Useful for occasional usage spikes.

Sovereignty & Compliance

On-premise No

Cloud only.

GDPR Compliant

Data residency: US

On-premise No

Cloud only.

This sheet's verification

Verified 19 June 2026

Pipecat STT Benchmark juin 2026 + Cartesia docs

Update note: Cartesia Ink-2 : WER 1.47%, $0.43/hr, 299ms TTFS (Pipecat STT Benchmark juin 2026). Nouvelle entrée juin 2026.

This date covers provider-specific prices, capabilities and notes. Comparative benchmarks follow the synchronization and methodology shown above.

ELO benchmarks / indices: Artificial Analysis