Cartesia vs Deepgram: pricing, features and which to choose

Cartesia is a San Francisco startup founded in 2023 by Karan Goel, Albert Gu, Arjun Desai, Brandon Yang and Chris Ré, researchers from Stanford's AI lab. Gu co-created the S4 and Mamba state space model architectures, which process long sequences with less memory than transformers, and that research is the basis of the company's products. Cartesia launched its Sonic text-to-speech model in 2024 and raised a $64 million Series A led by Kleiner Perkins in March 2025, bringing total funding to $91 million, according to Fortune. Its homepage names ServiceNow, Retell, Blue Machines and Quora among customers. The company's pitch is speed: models that start speaking quickly enough for live phone conversation, sold through one API that covers speech generation, transcription and complete voice agents.

Developers reach Cartesia through a REST and WebSocket API or the browser playground. Sonic turns text into streaming audio, with controls for speed, emotion and pronunciation, and supports more than 40 languages. Instant voice cloning creates a voice from a few seconds of audio on paid plans, and professional cloning trains a closer copy from longer recordings on the Startup plan and above. A localisation feature gives a cloned voice a new accent. Ink transcribes live audio for agents that need to hear callers. Managed Agents, the newest product, hosts complete phone agents with telephony, so a team can define an agent's prompt and tools without wiring a separate orchestration layer. Cartesia also offers on-premises and on-device deployment for customers with data residency rules.

Cartesia's voice library and creator tools are thinner than ElevenLabs, which also offers dubbing, sound effects and music, so it is a weaker fit for video producers and audiobook narrators. Managed Agents are newer than Vapi or Retell and have fewer integrations. Credit-based billing means long-form narration gets expensive faster on the low tiers. ElevenLabs Flash competes directly on latency with a larger voice catalog. Deepgram's Aura voices are cheaper per minute but less expressive, and Hume, once a rival on expressive speech, is closing its TTS API in November 2026. Pick Cartesia when you are building a real-time voice agent or interactive app and the delay before the first syllable is the metric you care about, or when you need on-device or on-premises speech models.

Deepgram is a San Francisco speech AI company founded in 2015 by Scott Stephenson and Noah Shutty, two physicists who had worked on dark matter detection, and it went through Y Combinator. It builds its own speech models instead of reselling others. In January 2026 it raised a $130 million Series C led by AVP at a $1.3 billion valuation, taking total funding past $215 million, and Stephenson told TechCrunch the company was cash-flow positive the year before. Investors include In-Q-Tel, Madrona, Tiger, Twilio and SAP. Deepgram says more than 1,300 organisations use its models, among them the meeting notetaker Granola, the voice agent platform Vapi and Twilio. Alongside the round it bought OfOne, a Y Combinator startup that builds drive-through ordering agents for quick-service restaurants.

Deepgram offers four API families. Speech-to-text covers pre-recorded files over REST and live audio over WebSockets, with speaker diarization, smart formatting, keyterm prompting for jargon and automatic language detection across 45-plus languages. Text-to-speech, sold under the Aura name, streams spoken replies for agents. The Voice Agent API bundles transcription, a language model and speech into one WebSocket so developers do not have to orchestrate three vendors. Audio Intelligence adds summaries, topic detection, sentiment and intent on top of transcripts. A typical build sends a phone or meeting audio stream to the API and receives word-level timestamps within a few hundred milliseconds. Developers get SDKs for Python, JavaScript, Go and .NET, plus a playground in the console for comparing models on their own audio.

Deepgram is an API for developers, so anyone who wants a finished meeting notetaker or dictation app should look at Otter.ai or similar products built on top of it. Self-serve support is limited to the community and Discord until you sign an enterprise contract. AssemblyAI is the closest rival, with comparable accuracy, strong audio intelligence features and similar per-hour pricing. Speechmatics is stronger on broad language coverage and on-premises deals in Europe, and OpenAI's speech models are convenient for teams already on the OpenAI API. ElevenLabs leads on expressive text-to-speech, where Deepgram's Aura voices are tuned for speed over range. Pick Deepgram when you need low-latency streaming transcription for agents or call analytics at high volume, especially if self-hosting is a requirement.