Deepgram
Deepgram is a voice AI infrastructure company providing real-time speech-to-text, text-to-speech, and conversational voice agent APIs built on proprietary end-to-end deep learning models. It serves enterprise contact-center, healthcare, financial services, and restaurant drive-thru customers, plus a developer base of 200,000+ users across cloud and self-hosted deployments.
- Company typePrivate
- Founded2015
- HeadquartersSan Francisco, United States
- Headcount251–500
- GTM typeB2B
- OfferingSoftware
What Deepgram does
Deepgram is a venture-backed voice AI infrastructure company founded in 2015 and headquartered in San Francisco, providing real-time speech-to-text, text-to-speech, and autonomous voice agent APIs. The platform is built on proprietary end-to-end deep learning architectures across three core model families: Nova-3 (multilingual STT across 50+ languages with sub-300ms latency), Flux (conversational speech recognition designed for voice agents, including a multilingual variant supporting real-time mid-call language switching across 10 languages), and Aura-2 (enterprise TTS with sub-200ms latency). These are unified through a Voice Agent API that bundles STT, LLM orchestration, and TTS into a single endpoint, and complemented by an Audio Intelligence add-on suite for entity recognition, summarization, sentiment, and intent extraction.
The company's technology supports both cloud and self-hosted deployment, with confidential-computing configurations for regulated industries. Deepgram serves enterprise customers across contact centers, healthcare, financial services, and the quick-service restaurant vertical (via the OfOne acquisition forming Deepgram for Restaurants), plus a long-tail developer base of more than 200,000 users and 1,300+ organizations. Marquee integrations include IBM watsonx Orchestrate (as IBM's first voice partner), Amazon SageMaker and Amazon Connect, Twilio, Cloudflare, Genesys, and LiveKit. Distribution combines a self-serve product-led growth funnel (free $200 credits, console.deepgram.com) with enterprise direct sales and OEM/embedded partnerships.
Deepgram's revenue model is usage-based for self-serve and consumption-based API workloads (STT, TTS, voice agent minutes/characters), supplemented by quote-based multi-year enterprise contracts that include dedicated support, self-hosted deployment, and volume pricing. As of January 2026 the company is cash-flow positive, employs 101-250 people distributed across 20+ US states and 5 countries, and operates regional offices in Singapore (APAC) and London (EMEA). The company raised a $130M Series C at a $1.3B valuation in January 2026, bringing total disclosed funding above $215M.
Deepgram firmographics
Firmographics- Name
- Deepgram
- Legal name
- Deepgram Inc.
- Website
- https://deepgram.com
- Company type
- Private
- Founded year
- 2015
- Operating status
- Operating
- Headcount range
- 251–500 employees
- Short description
- Deepgram is a voice AI infrastructure company providing real-time speech-to-text, text-to-speech, and conversational voice agent APIs built on proprietary end-to-end deep learning models. It serves enterprise contact-center, healthcare, financial services, and restaurant drive-thru customers, plus a developer base of 200,000+ users across cloud and self-hosted deployments.
- Ownership category
- akta.pro rank
Deepgram industry classification
Industry- Product category
- Voice AI and Speech Recognition Software
- NAICS
- Computer Systems Design and Related Services (54151), Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (518)
- SIC
- Services-Computer Programming, Data Processing, Etc. (7370), Communications Services, Nec (4899)
- akta.pro primary industry
- Contact Center Speech AI (agent assist, IVR automation) (HDAAAFAL)
- akta.pro secondary industries
- Audio & Speech Analytics (call analytics, QA, insights) (HDAAAFAH), Speech Translation & Multilingual Speech Tech (HDAAAFAE), Multimodal Generation (Text-Image-Video-Audio) & Creative Toolchains (HDAAACAG)
Keywords
Where Deepgram is headquartered
LocationHeadquarters
- HQ city
- San Francisco
- HQ country
- United States
- HQ region
- North America
Offices3 records
Markets served
Deepgram business model
Business model- GTM type
- B2B
- Offering type
- Software
- Cost components
- Technology or R&D, Personnel, Infrastructure, Marketing or Sales, Operations
Revenue model
- API Usage-Based Pricing: Consumption-based pricing for speech-to-text, text-to-speech, and voice agent API calls. Customers pay per minute of audio processed or per character of text synthesized. Starts at $0.0052/min for pre-recorded audio.
- Enterprise Contracts: Custom enterprise agreements for large-scale deployments with dedicated infrastructure, support, and volume discounts. Includes self-hosted deployment options for regulated industries.
Pricing tiers
| Model | Billing | Price |
|---|---|---|
| Freemium | Pay-as-you-go | Free trial tier with $200 credits |
| Usage-based | Pay-as-you-go | Usage-based pricing starting at $0.0052/min |
| Subscription | Multi-year contract | Enterprise custom pricing |
Go-to-market motion3 records
Distribution channels6 records
Marketing channels5 records
Deepgram product offering
Product offeringCore offering
Deepgram provides an end-to-end voice AI platform delivered through APIs for real-time speech-to-text (Nova-3), text-to-speech (Aura-2), conversational speech recognition (Flux), and unified voice agent orchestration. The platform supports 50+ languages and serves developers, contact centers, healthcare, financial services, and quick-service restaurants through cloud, self-hosted, and OEM-embedded deployments.
Product overview
Deepgram is a foundational voice AI company offering a unified real-time Voice AI platform providing speech-to-text (STT), text-to-speech (TTS), and autonomous voice agent capabilities. The core product portfolio centers on Nova-3 (speech-to-text), Aura-2 (text-to-speech), and Flux (conversational speech recognition for voice agents), unified through the Voice Agent API. These are complemented by Audio Intelligence add-on modules for extracting insights from transcribed audio. Deepgram offers multiple deployment options including cloud API, Amazon SageMaker integration, and self-hosted on-premises deployment for regulated industries. The company operates a vertical-specific offering, Deepgram for Restaurants, acquired via OfOne for drive-thru automation with 95%+ containment rates. Developer support includes SDKs in Python, JavaScript, .NET, Go, and Java, plus a Startup Program for early-stage companies.
Differentiator
Problem solved
Functional benefit
Brands
- Deepgram for Restaurants: AI voice automation platform for quick-service restaurants and drive-thru operations, formed from the OfOne acquisition
- Nova
- Aura
- Flux
Products and services
- Nova-3 Speech-to-Text Flagship automatic speech recognition (ASR) model offering sub-300ms latency for real-time streaming and high-accuracy transcription for pre-recorded audio. Supports 50+ languages with multilingual and code-switching capabilities. Used by developers and enterprises building voice AI applications requiring fast, accurate speech-to-text conversion.
- Aura-2 Text-to-Speech Enterprise-grade text-to-speech (TTS) model delivering sub-200ms latency for real-time applications and domain-specific pronunciation for structured data. Supports flexible deployment across cloud, VPC, and on-premises environments. Used by developers building voice agents and conversational AI applications requiring natural-sounding synthesized speech.
- Flux Conversational Speech Recognition Conversational speech recognition model designed for voice agent applications with turn detection under 400ms. Flux Multilingual variant supports real-time language detection and mid-call switching across 10 languages within a single conversation. Used by developers building real-time voice agents requiring natural dialogue flow and interruption handling.
- Audio Intelligence API
Quantifiable outcome
- Sub-200ms latency for text-to-speech responses
- +3 more outcomes
Companies that use Deepgram
Customer profileNamed customers11 records
Segments5 records
Ideal customer profiles5 records
Deepgram technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- Yes
- API docs
- API detail
Core technology
AI maturity
App detail
Integration15 records
AI capability7 records
Feature6 records
Deepgram partnerships and signals
Strategic signalPartnerships
Six partnerships are on record, tiered core.
- Penguin SolutionscorePenguin Solutions became an NVIDIA AI Factory Specialized Partner, citing existing collaborations with Deepgram for AI inference platforms. Partnership includes deployment of the Haein AI Factory with SK Telecom.
- FortanixcorePartnership with Fortanix and NVIDIA to enable fully encrypted, on-premises voice AI deployments for regulated industries including healthcare, finance, and government. Leverages NVIDIA Confidential Computing-enabled GPUs and Fortanix Confidential AI platform.
- Penguin SolutionscorePenguin Solutions deployed optimized AI inference infrastructure for Deepgram's enterprise voice AI capabilities using Dell PowerEdge servers with NVIDIA RTX PRO 6000 Blackwell GPUs and Dell PowerScale storage. Partnership targets healthcare and retail sectors.
- IBMcoreIBM integrated Deepgram's speech-to-text and text-to-speech technology into watsonx Orchestrate, making Deepgram IBM's first voice partner. The partnership enables enterprise customers to build voice agents and voice-driven workflows with real-time transcription supporting dozens of Arabic and Indian dialects.
- OfOnecoreDeepgram acquired OfOne, an AI-native voice platform for restaurants and quick-service drive-thru markets. The acquisition forms the foundation of Deepgram for Restaurants offering.
- Amazon Web Services (AWS)coreDeepgram announced native integration with Amazon SageMaker and Amazon Connect, enabling enterprises to deploy streaming real-time speech-to-text, text-to-speech, and voice agent capabilities. Deepgram operates as an AWS Generative AI Competency Partner under a multi-year Strategic Collaboration Agreement.
Scale indicators8 records
Recent moves6 records
Expansion highlights7 records
Deepgram competitors and assessment
Company assessmentBroad incumbents
- Amazon Web Services (Amazon Transcribe / Polly / Bedrock): AWS offers Amazon Transcribe, Polly, and Bedrock-based voice AI, competing with Deepgram across price, scale, and bundling, and is simultaneously a Deepgram distribution partner via SageMaker and Amazon Connect, creating an ambiguous partner-competitor dynamic.
- Microsoft Azure Speech: Azure AI Speech provides STT, TTS, and speech translation as part of Microsoft's cloud stack, competing with Deepgram for enterprise customers and offering similar multilingual and real-time capabilities, often bundled into enterprise agreements.
- Cerence: Cerence is a leading provider of AI-powered automotive voice assistants, comparable to Deepgram in conversational speech recognition, multilingual support, and embedded voice AI, though specialized in the automotive vertical.
- OpenAI (Whisper / Realtime API): OpenAI offers Whisper for STT and the Realtime API combining STT/TTS with GPT-4o, providing a foundation-model alternative that competes with Deepgram's core API business and could disintermediate dedicated voice AI infrastructure.
- Google Cloud Speech-to-Text / Text-to-Speech: Google Cloud offers STT and TTS as part of its broader cloud platform, directly competing with Deepgram on accuracy, multilingual coverage, and pricing, with the advantage of deep cloud bundling and enterprise channel.
Direct peers
- ElevenLabs: ElevenLabs is a leading TTS and voice generation platform, directly competing with Deepgram's Aura-2 product in enterprise voice synthesis, voice cloning, and conversational AI use cases.
- Rev.ai: Rev.ai offers ASR and speech understanding APIs with a developer-focused model, directly competing with Deepgram in transcription accuracy, real-time streaming, and enterprise transcription use cases.
- AssemblyAI: AssemblyAI is a direct STT/TTS/AI API competitor offering speech-to-text, speech understanding, and audio intelligence models via API to developers and enterprises, with a very similar product-led growth and per-minute pricing model to Deepgram.
- Speechmatics: Speechmatics provides enterprise-grade automatic speech recognition with strong multilingual support, including real-time and batch transcription, directly competing with Deepgram in regulated-industry and global enterprise deployments.
Emerging players
- SoundHound: SoundHound provides voice AI for automotive, restaurant drive-thru, and customer service, partially overlapping with Deepgram in QSR and contact-center verticals following the OfOne acquisition.
Market position
Strengths5 records
Weaknesses5 records
Competitive moat6 records
Key risks6 records
Key highlights7 records
Customer concentration
Deepgram social profiles
Digital presenceDeepgram compliance and trust
Trust signalCompliance4 records
Deepgram financial estimates
Financial estimateRevenue estimate
Valuation estimate
Deepgram leadership team
Management profileNumber of profiles
Profiles6 records
Deepgram subsidiaries and ownership
Company hierarchySubsidiaries1 record
Deepgram funding detail
Funding detailFunding overview
Funding rounds12 records
Investors33 records
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
Deepgram M&A and investment
M&A and investmentM&A2 records
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about Deepgram
What does Deepgram do?
Deepgram provides an end-to-end voice AI platform delivered through APIs for real-time speech-to-text (Nova-3), text-to-speech (Aura-2), conversational speech recognition (Flux), and unified voice agent orchestration. The platform supports 50+ languages and serves developers, contact centers, healthcare, financial services, and quick-service restaurants through cloud, self-hosted, and OEM-embedded deployments.
Is Deepgram a public or private company?
Deepgram is a private company. It is classified as venture growth investor backed and is currently operating.
When was Deepgram founded?
Deepgram was founded in 2015. It employs 251 to 500 people.
Where is Deepgram based?
Deepgram is headquartered in San Francisco, United States, in the North America region.
How does Deepgram make money?
Two revenue lines are on record. API Usage-Based Pricing is the primary driver. The others are enterprise Contracts.
Who are Deepgram's main competitors?
Broad incumbents on record are Amazon Web Services (Amazon Transcribe / Polly / Bedrock), Microsoft Azure Speech, Cerence, OpenAI (Whisper / Realtime API) and Google Cloud Speech-to-Text / Text-to-Speech. Direct peers are ElevenLabs, Rev.ai, AssemblyAI and Speechmatics. SoundHound is listed as an emerging player.
Does Deepgram have an API?
Yes. Deepgram offers a public API platform providing real-time speech-to-text (STT), text-to-speech (TTS), and voice agent capabilities. The API supports WebSocket-based streaming for real-time applications and REST/POST for pre-recorded audio. Authentication uses token-based API keys. The platform offers SDKs in multiple languages including Python, JavaScript, .NET, Go, and Java. Free tier includes $200 in credit. Self-hosted deployment options are available for enterprise customers requiring on-premises infrastructure. Developer documentation is at developers.deepgram.com.
What industry is Deepgram in?
Deepgram's product category is Voice AI and Speech Recognition Software. Its primary akta.pro industry code is HDAAAFAL, Contact Center Speech AI (agent assist, IVR automation), with a secondary code of HDAAAFAH, Audio & Speech Analytics (call analytics, QA, insights). Its NAICS code is 54151 and its SIC code is 7370.