LMNT
LMNT is an AI-native text-to-speech platform offering fast, lifelike speech synthesis via REST and WebSocket APIs. Serving developers building voice agents, conversational AI, gaming, and content localization applications across 31 languages with voice cloning and sub-200ms latency.
- Company typePrivate
- Founded2019
- HeadquartersCalifornia, United States
- Headcount11–50
- GTM typeB2B
- OfferingSoftware
What LMNT does
LMNT is an AI-native text-to-speech platform built around its proprietary Blizzard model family (currently Blizzard 2.0). The company offers two primary API surfaces: a REST-based Speech API for pre-produced voiceovers and audio content, and a WebSocket-based Speech Sessions API that streams text from large language models into real-time synthesized speech with 150-200ms time-to-first-audio latency. The platform supports 31 languages with native code-switching, voice cloning from 5-10 seconds of reference audio, accent control, and word-level timestamps, serving developers building voice agents, conversational AI, content localization, gaming, and educational applications.
LMNT monetizes through tiered self-serve subscriptions (Free, Indie at $10/month, Pro at $49/month, Premium at $199/month) with usage-based overage charges down to $0.035 per 1,000 characters, plus custom enterprise plans for high-volume customers. Go-to-market is API-first and product-led, anchored by a free Playground, official Python/TypeScript/Go SDKs, and a Startup Grant program offering 45 million free characters over 3 months to qualifying early-stage voice AI startups. Distribution is amplified through integrations with Vercel, LiveKit, Pipecat, Vapi, Claude Code, Cursor, and Augment Code, embedding LMNT into the dominant voice agent and agentic coding ecosystems. Named customers include Khan Academy, HeyGen, Vapi, Fixie, Vercel, Unity, Replit, and Pipecat, spanning education, video creation, voice AI, developer tooling, and gaming.
The company is privately held, headquartered in Palo Alto, California, employs 11-50 people, and is backed by Elad Gil and Conviction. Founders Sharvil Nanavati and Zach Johnson lead a small team with prior experience at Google[x], Meta, and Microsoft. The company holds SOC-2 Type II certification, indicating enterprise-grade security posture appropriate for handling customer voice content.
LMNT firmographics
Firmographics- Name
- LMNT
- Legal name
- LMNT, Inc.
- Website
- https://lmnt.com
- Company type
- Private
- Founded year
- 2019
- Operating status
- Operating
- Headcount range
- 11–50 employees
- Short description
- LMNT is an AI-native text-to-speech platform offering fast, lifelike speech synthesis via REST and WebSocket APIs. Serving developers building voice agents, conversational AI, gaming, and content localization applications across 31 languages with voice cloning and sub-200ms latency.
- Ownership category
- akta.pro rank
LMNT industry classification
Industry- Product category
- Text-to-Speech API
- NAICS
- Software Publishers (513210), Software Publishers (5132)
- SIC
- Services-Prepackaged Software (7372)
- akta.pro primary industry
- Voice Cloning, Dubbing & Audio Generative Media (HDAAAFAK)
- akta.pro secondary industries
- Audio & Speech Analytics (call analytics, QA, insights) (HDAAAFAH), Speech Translation & Multilingual Speech Tech (HDAAAFAE)
Keywords
Where LMNT is headquartered
LocationHeadquarters
- HQ city
- California
- HQ country
- United States
- HQ region
- North America
Offices1 record
Markets served
LMNT business model
Business model- GTM type
- B2B
- Offering type
- Software
- Cost components
- Personnel, Technology or R&D, Infrastructure, Marketing or Sales
Revenue model
- API Usage-based Subscription: LMNT sells tiered subscription plans (Free, Indie, Pro, Premium) that include monthly character allowances. Plans are billed monthly (Annual billing available with discounts). Usage beyond the included allowance is charged per 1,000 characters. All plans include unlimited voice clones and commercial licensing. Enterprise/custom plans are available for high-volume users.
- Overage Charges: When users exceed their monthly character limit, they are charged per 1,000 characters at tier-specific rates (e.g., $0.05/1K for Indie, $0.045/1K for Pro, $0.035/1K for Premium). This creates usage-based upsell dynamics as customers scale.
- Startup Grant Program: Eligible startups (fewer than 20 employees, building voice AI products) receive 45M free credits over 3 months (15M characters/month). After 3 months, customers are automatically upgraded to an enterprise plan or can adjust their plan.
Pricing tiers
| Model | Billing | Price |
|---|---|---|
| Freemium | Monthly | Free tier for trying out the product |
| Subscription | Monthly | Entry-level paid plan for individual developers and small projects |
| Subscription | Monthly | Mid-tier plan for professionals and growing teams |
| Subscription | Monthly | High-volume plan for production workloads |
| Freemium | Monthly | Startup Grant: 3 months free for eligible early-stage startups |
Go-to-market motion2 records
Distribution channels3 records
Marketing channels8 records
LMNT product offering
Product offeringCore offering
LMNT provides an AI text-to-speech platform powered by its proprietary Blizzard speech models, delivered through a REST Speech API and a WebSocket Speech Sessions API for real-time streaming. The platform generates lifelike speech in 31 languages with native code-switching, supports voice cloning from 5-10 seconds of reference audio, accent control, and word-level timestamps, targeting developers and businesses building voice agents, conversational AI, content localization, and gaming applications.
Product overview
LMNT is an AI speech technology company offering a unified text-to-speech platform centered on its proprietary Blizzard model family. The core product is the LMNT Text-to-Speech API, which includes two primary interfaces: the Speech API for pre-produced text-to-speech (voiceovers, audiobooks, localization) and the Speech Sessions API for real-time streaming with LLMs (voice agents, conversational AI). Supporting both is Voice Cloning, enabling custom voices from 5-second samples. The platform is accessible via REST API and WebSocket, with official SDKs for Python, TypeScript, and Go. Pricing tiers (Free, Indie at $10/mo, Pro at $49/mo, Premium at $199/mo) include varying character limits with a Startup Grant program offering 45M free credits for qualifying early-stage companies. The Blizzard model receives regular updates (Blizzard 1.0 through 2.0) with expanded language support and quality improvements.
Differentiator
Problem solved
Functional benefit
Brands
- Blizzard: Current major model family supporting 31 languages, voice cloning, accent control, word timestamps, and streaming/speech sessions.
- Blizzard 2.0
Products and services
- LMNT Text-to-Speech API RESTful API at api.lmnt.com that converts text into natural, lifelike speech with support for 31 languages, voice cloning, low-latency streaming, and word timestamps for synchronization. Built for developers and businesses integrating speech synthesis into their applications.
- Speech Sessions API WebSocket-based API that streams text progressively from an LLM and returns generated speech in real time, designed for voice agents, conversational AI, and latency-sensitive applications with 150-200ms time-to-first-audio.
- Voice Cloning Capability to create custom voices from 5-10 seconds of reference speech, available on all plans with unlimited voice clones and instant clone quality meeting professional standards.
- Blizzard Speech Model Proprietary speech model family (currently Blizzard 2.0) optimized for low-latency speech synthesis, voice cloning, accent control, and 31-language support with native code-switching. Updated regularly since the initial Blizzard preview in December 2024.
- Playground Free browser-based interface at app.lmnt.com that lets users generate speech, clone voices, and experiment with LMNT's API before writing code.
Quantifiable outcome
- Latency as low as 150–200ms for low-latency streaming
- +3 more outcomes
Companies that use LMNT
Customer profileNamed customers8 records
Segments4 records
Ideal customer profiles3 records
LMNT technology and API
TechnologyTechnology focussed Yes
API detail
- Has API
- Yes
- API docs
- API detail
Core technology
AI maturity
App detail
Integration10 records
AI capability3 records
Feature6 records
LMNT partnerships and signals
Strategic signalPartnerships
Four partnerships are on record, tiered core.
- VercelcoreLMNT has an official integration listed in the Vercel marketplace. Vercel users can install the LMNT integration to add LMNT_API_KEY environment variables to their projects with one-click deployment. The integration supports Python, TypeScript, Go, and Ruby runtimes on Vercel.
- LiveKitcoreLiveKit is listed as a supported integration for building realtime speech experiences. The Speech Sessions API is designed to work with LiveKit for real-time voice agent applications, combining WebRTC infrastructure with LMNT's speech streaming.
- PipecatcorePipecat is a voice AI framework partner that builds conversational AI products using LMNT's speech APIs. Pipecat's platform enables developers to create voice agents that leverage LMNT's Speech Sessions API.
- VapicoreVapi is a voice AI platform and customer of LMNT, building telephony and voice agent solutions using LMNT's speech API. Vapi is listed as a trusted customer logo on LMNT's website.
Scale indicators5 records
Recent moves6 records
Expansion highlights5 records
LMNT competitors and assessment
Company assessmentDirect peers
- PlayHT: API-first TTS platform with voice cloning and a large voice library. Competes with LMNT on developer-grade speech synthesis, multilingual coverage, and usage-based pricing tiers. Targets content creators, developers, and enterprises.
- Cartesia: Real-time voice AI platform focused on low-latency speech synthesis and voice agents. Comparable to LMNT's Speech Sessions API use case and voice-agent developer target market. Emphasizes on-device/inference speed similar to LMNT's latency positioning.
- WellSaid Labs: API-based TTS with voice cloning targeted at enterprise content production. Comparable to LMNT on commercial-grade voice synthesis, though typically positioned more toward studio/voiceover use cases than real-time conversational AI.
- ElevenLabs: Direct competitor offering API-based TTS with voice cloning, multilingual support, and conversational AI products. Comparable in product category, target market (developers and enterprise), and pricing model. Most analogous peer to LMNT in the AI speech synthesis space.
- Resemble AI: Voice cloning and TTS platform with enterprise focus, real-time synthesis, and deepfake detection. Competes with LMNT on voice cloning quality and enterprise speech synthesis use cases, including conversational AI applications.
Emerging players
- Hume AI: Voice AI company with emotionally-aware TTS (Octave) and conversational voice agents. Overlaps with LMNT in real-time voice synthesis for AI agents, with an additional focus on emotional expression that LMNT addresses through voice prompting.
- Murf AI: TTS platform with strong focus on content creators, voiceovers, and enterprise localization. Overlaps with LMNT on multilingual voice generation and natural-sounding speech but skews more toward studio/content workflows than real-time voice agents.
Broad incumbents
- Google Cloud Text-to-Speech: Cloud TTS service bundled into Google Cloud Platform with broad language support. Comparable to LMNT on multilingual speech but lacks the developer-led voice cloning and code-switching capabilities that differentiate LMNT.
- OpenAI (TTS / Voice Engine): Offers TTS APIs (tts-1, tts-1-hd, Voice Engine) as part of a much broader AI platform. Overlaps directly with LMNT's product category but bundles TTS with LLMs, giving it distribution advantages and bundling economics that LMNT must compete against.
- Amazon Polly: Long-established cloud TTS service bundled into AWS. Competes with LMNT on API-based speech synthesis, particularly for enterprise customers already on AWS, though typically lacks LMNT's voice cloning quality and low-latency conversational focus.
Market position
Strengths5 records
Weaknesses5 records
Competitive moat6 records
Key risks6 records
Key highlights7 records
Customer concentration
LMNT social profiles
Digital presenceLMNT compliance and trust
Trust signalCompliance1 record
LMNT financial estimates
Financial estimateRevenue estimate
Valuation estimate
LMNT leadership team
Management profileNumber of profiles
Profiles2 records
LMNT funding detail
Funding detailFunding overview
Funding rounds
Investors
Funding detail is available on the Subscription and Enterprise plan.Contact sales →
LMNT M&A and investment
M&A and investmentM&A
Investments
M&A and investment is available on the Subscription and Enterprise plan.Contact sales →
Frequently asked questions about LMNT
What does LMNT do?
LMNT provides an AI text-to-speech platform powered by its proprietary Blizzard speech models, delivered through a REST Speech API and a WebSocket Speech Sessions API for real-time streaming. The platform generates lifelike speech in 31 languages with native code-switching, supports voice cloning from 5-10 seconds of reference audio, accent control, and word-level timestamps, targeting developers and businesses building voice agents, conversational AI, content localization, and gaming applications.
Is LMNT a public or private company?
LMNT is a private company. It is classified as venture growth investor backed and is currently operating.
When was LMNT founded?
LMNT was founded in 2019. It employs 11 to 50 people.
Where is LMNT based?
LMNT is headquartered in California, United States, in the North America region.
How does LMNT make money?
Three revenue lines are on record. API Usage-based Subscription is the primary driver. The others are overage Charges and startup Grant Program.
Who are LMNT's main competitors?
Direct peers on record are PlayHT, Cartesia, WellSaid Labs, ElevenLabs and Resemble AI. Emerging players are Hume AI and Murf AI. Broad incumbents are Google Cloud Text-to-Speech, OpenAI (TTS / Voice Engine) and Amazon Polly.
Does LMNT have an API?
Yes. RESTful API at https://api.lmnt.com providing programmatic access to LMNT's text-to-speech and voice cloning models. Includes Speech API (generate speech from text with optional word-level timestamps), Speech Sessions API (stream text progressively and receive speech in real-time via WebSocket), Voices API (create, list, retrieve, update, and delete voices), and Accounts API (usage and plan information). Authentication via X-API-Key header and lmnt-version header required. No concurrency or rate limits on paid plans. Available at https://api.lmnt.com with documentation at https://docs.lmnt.com. Developer documentation is at docs.lmnt.com.
What industry is LMNT in?
LMNT's product category is Text-to-Speech API. Its primary akta.pro industry code is HDAAAFAK, Voice Cloning, Dubbing & Audio Generative Media, with a secondary code of HDAAAFAH, Audio & Speech Analytics (call analytics, QA, insights). Its NAICS code is 513210 and its SIC code is 7372.