ElevenLabs

Name: ElevenLabs
Brand: ElevenLabs
Rating: 4 (1834 reviews)

AI NativeLeader

Transforming how we interact with technology through AI research and products.

Est 2022|Valuation $11B|Raised $850.4M|160 Ref Customers

LinkedIn 384K|Website|Web traffic 58.2M|FTEs 800|HQ New York, NY

MarketsAI Voice Agent Platform Voice AI Agents

Published Oct 15, 2026

01Common complaints

- Current limitations acknowledged/observed • v2 Realtime integration requires WebSockets; existing SDKs/apps may need refactoring. • Occasional status incidents causing 429/500s or regional outages. • Users reported diarization inconsistencies via API and export UI bugs (April 2025). - Roadmap signals (2025) • Scribe v2 Realtime shipped November 2025 following the v1 “real‑time soon” promise; docs and cookbooks updated. • HIPAA enablement via ZRM; diarization length increases (up to 2 hours). Sources: docs; changelog; launch posts; status updates. - Scale limitations • STT concurrency caps by plan; Enterprise can elevate. Build back‑pressure and retry logic into clients. - Security/compliance • ZRM disables logging/storage; customers must capture outputs via webhooks for audit. • BAA available for qualifying customers; data residency offered (EU/India isolated environments). Sources: Trust Center; ZRM; GDPR/data residency docs. - Architectural/technical debt indicators • UI/API mismatch bugs (e.g., dubbing flag) and rollout naming confusion for Scribe v2 suggest rapid iteration; expect occasional documentation drift.

02Why they buy

- Indications from customer/developer feedback • Reported multilingual accuracy and performance in noisy/complex scenarios; developer anecdotes and internal/press benchmarks highlight Scribe relative to low‑latency peers. • Enterprise privacy features: Zero Retention Mode; SOC 2/ISO; HIPAA with BAA; EU/India data residency. - Features/capabilities • Predictive “negative latency” behavior (next‑word/punctuation prediction) to reduce effective latency; manual commit control; VAD; flexible audio encodings. - Differentiation vs competitors • Combines accuracy, latency, and compliance requirements in one stack and integrates with ElevenLabs’ TTS/dubbing/Agents. - Where competitors have clear advantages • Pricing (OpenAI/AssemblyAI); on‑prem deployments (AssemblyAI/Speechmatics); third‑party ASR feature breadth (varies by vendor). - Market perception vs reality • Marketing claims of “most accurate” are supported on selected multilingual benchmarks; independent evaluations show mixed leaders depending on dataset and formatting rules. Buyers should run narrow, representative A/B tests.

• Enterprise privacy features: Zero Retention Mode; SOC 2/ISO; HIPAA with BAA; EU/India data residency.

Key Differentiator

03Field reports

Note: Scribe‑specific, third‑party reviews remain limited because Scribe v2 Realtime launched November 2025. The quotes below include ElevenLabs platform users (TTS/Agents/STT) plus developer discussions and early Scribe feedback. Positive - “The voice quality is outstanding, natural, and expressive… the platform is fast, clean, and easy to use.” — G2 review, Oct 29, 2025. Source: G2. - “Upgrading to Scribe significantly improved our product quality… speed tripled — a one‑hour meeting is now processed in 30–45 seconds.” — Jamie (customer story). Source: ElevenLabs blog customer story. - “Results I get are amazingly good… already were excellent with V1.” — Reddit user discussing Scribe V2 vs V1. Source: Reddit. Critical/Negative - “You are charged for every single generation… cost becomes astronomical.” — Trustpilot reviewer (audiobook workflow; billing policy frustration). Source: Trustpilot. - “Export function doesn’t seem to work… click ‘Export,’ nothing happens.” — Reddit (Speech‑to‑Text export bug report, Apr 2025). Source: Reddit. - “The menu item disappeared… couldn’t use transcription credits… turned out to be ad‑blocker.” — Reddit (dashboard STT UI issue; later resolved by whitelisting). Source: Reddit. - “As Scribe v2 is only real‑time… requires WebSockets.” — Reddit dev thread on integration friction. Source: Reddit. - “API… unable to capture diarization… only Speaker 0 throughout the file.” — Reddit (Scribe v1 diarization complaint). Source: Reddit. Why buyers chose ElevenLabs/Scribe - Accuracy: internal/third‑party write‑ups cite lower WER vs competing low‑latency models; developers report strong multilingual performance. Sources: ElevenLabs research post; Daily AI Primer; VentureBeat; independent blogs. - Enterprise posture: SOC 2/ISO 27001; HIPAA with BAA; Zero Retention Mode; EU/India data residency. Sources: Trust Center; docs. Why some left/hesitated - Pricing/billing misunderstandings or perceived expense in iterative workflows. Sources: Trustpilot; G2 snippets; Reddit. - Occasional outages/429s in STT (load spikes; region GPU capacity). Sources: Status page incidents. - Feature gaps/bugs (export issues, diarization inconsistencies, UI/API mismatches; docs confusion about model IDs). Sources: Reddit threads, docs. Specific outcomes - Jamie (meeting assistant): 3× faster pipeline; speaker errors reduced; multilingual performance. Source: ElevenLabs blog customer story. - Early developer reports testing Realtime V2 report “amazingly good” results relative to V1 but note integration changes for WebSockets. Source: Reddit. Deal‑breakers mentioned - Strict on‑device or fully offline requirements (privacy or air‑gapped constraints) — cloud‑hosted Scribe is unsuitable. - Alternative: open‑source Whisper or Speechmatics on‑prem. Sources: AssemblyAI (self‑hosted options), developer discussions. 1) Jamie (meeting assistant SaaS) - Context: Switched from a complex custom diarization/transcription pipeline to Scribe. - Results: 3× faster transcription (1h meeting → 30–45 s); complaints about speaker errors “disappeared”; multilingual coverage improved (EN/DE/ES/NL). - Implementation: “Few days” to integrate; minimal customization; reduced engineering overhead. - Challenges: Prior pipeline complexity; Scribe addressed overlapping speech and non‑verbal audio event tagging. Source: ElevenLabs customer story. 2) Developer pilots (mixed) - Positive: “Amazingly good” transcriptions vs V1; language detection and accuracy strong. - Friction: Realtime requires WebSockets; incorrect parameter usage led to defaulting PCM 16 kHz; documentation/model naming confusion during rollout. Sources: Reddit threads; realtime docs. 3) Operational reliability (partial outages) - Incidents: STT increased heavy‑load errors; EU data residency STT outage due to GPU availability; incidents resolved same day but highlight capacity sensitivity. Source: ElevenLabs status pages. Note: Independent, metric‑rich public case studies specific to Scribe v2 Realtime are limited as of Nov 26, 2025.

04Best fit

- Company sizes: Startups building agents/notetakers; SMBs running multilingual support desks; mid‑market/enterprise voice operations (support/sales) where latency and compliance matter. - Industries/verticals: • Contact centers and CX tech (real‑time agent assist). • SaaS meeting intelligence tools. • Media/streaming and events (live captions). • Healthcare (with BAA and ZRM) for ambient notes/transcription (sensitive data). - Less ideal segments: • Air‑gapped/on‑premise only environments. • Extremely cost‑constrained batch workloads where $/min overrides accuracy/controls. - Roles/personas: Developer leads/VP Eng, Head of AI/ML, Product Managers building agents/notetakers, IT/Sec reviewing compliance (SOC 2/ISO/HIPAA, data residency). - Pain points addressed: Latency (talk‑over); multilingual accuracy; speaker diarization; enterprise privacy controls. - Qualification criteria: Need sub‑200 ms turns; multilingual; desire EU/India residency or ZRM; willing to use WebSockets/streaming. - Deal disqualifiers: Must be on‑device; strict no‑cloud; require custom on‑prem model training; require guaranteed 0% downtime (cloud dependency). - Real customer example: Jamie (meeting assistant) — faster activation/usage tied to tripled processing speed. Source: ElevenLabs case study.

• Extremely cost‑constrained batch workloads where $/min overrides accuracy/controls.

Key Differentiator

05The pitch

- What it is: Scribe is ElevenLabs’ automatic speech recognition (ASR) line. In 2025, the company launched Scribe v1 (batch STT across 99 languages) and, on November 11, 2025, introduced Scribe v2 Realtime (“Scribe2”) for low‑latency streaming ASR with ~150 ms end‑to‑end latency across 90+ languages. Product is available via API and inside ElevenLabs’ Agents platform. Sources: elevenlabs.io blog/product pages; docs; press coverage. - Core value proposition: High‑accuracy multilingual transcription with enterprise controls. Scribe v2 Realtime targets voice agents, live captioning, call‑center assistants, and meeting assistants where latency affects user experience. Sources: product pages, docs, launch blogs. - Target market and positioning: • Builders of real‑time conversational agents (customer support, sales, in‑product voice UX). • SaaS “AI notetaker” and meeting intelligence apps. • Media/live events needing live captions and multilingual support. • Enterprises requiring SOC 2/ISO 27001/HIPAA options, Zero Retention Mode, and EU/India data residency. Sources: launch posts; trust center; docs. - Company background and funding: ElevenLabs (founded 2022; London/NYC) is known for TTS/voice cloning and dubbing. Series C ($180M) closed January 30, 2025 at a $3.3B valuation; in November 2025 the company announced a $6.6B employee tender valuation and reported >$200M ARR. Sources: Reuters; TechCrunch; ElevenLabs blog; Business Insider. - Pricing model and cost considerations: • Scribe v1 batch STT is included in subscription plans measured in “hours” (e.g., Starter $5/mo includes 12h 30m; Business $1,320/mo includes 6,000 hours; overages vary). • Earlier public communications also cited $0.40/hour list pricing for Scribe v1. • For real‑time (Scribe v2 Realtime), pricing is presented under ElevenLabs Agents/Realtime docs; TTS/Agents minutes appear on the broader pricing page. • Concurrency limits for STT are higher than other ElevenLabs services (e.g., 20 on Creator, 60 on Business; Enterprise elevated). • Note: conversational AI usage adds non‑STT costs (LLM/tooling), and ElevenLabs has adjusted voice‑agent pricing over 2025. Sources: Scribe docs/pricing pages; blog on pricing changes; TechCrunch coverage. - Key competitors • Deepgram (Nova‑3, Flux) — reports of real‑time accuracy and speed, per‑second billing. • OpenAI (gpt‑4o[-mini] transcribe, Whisper) — low pricing relative to peers; ecosystem integrations. • AssemblyAI (Universal‑Streaming) — low list price, states unlimited concurrency, self‑hosted/VPC options. • Google Cloud Speech‑to‑Text v2, AWS Transcribe, Azure AI Speech — hyperscaler options. • Speechmatics, Rev.ai, Gladia — enterprise/vertical plays. Sources: competitor pricing pages and comparative articles.

Get a comprehensive analysis of ElevenLabs including market position, competitive landscape, adoption trends, and peer benchmarks.

Download full report →

Published Oct 15, 2026

01Common complaints

Current limitations acknowledged/observed
• v2 Realtime integration requires WebSockets; existing SDKs/apps may need refactoring.
• Occasional status incidents causing 429/500s or regional outages.
• Users reported diarization inconsistencies via API and export UI bugs (April 2025).
Roadmap signals (2025)
• Scribe v2 Realtime shipped November 2025 following the v1 “real‑time soon” promise; docs and cookbooks updated.
• HIPAA enablement via ZRM; diarization length increases (up to 2 hours).
Sources: docs; changelog; launch posts; status updates.
Scale limitations
• STT concurrency caps by plan; Enterprise can elevate. Build back‑pressure and retry logic into clients.
Security/compliance
• ZRM disables logging/storage; customers must capture outputs via webhooks for audit.
• BAA available for qualifying customers; data residency offered (EU/India isolated environments).
Sources: Trust Center; ZRM; GDPR/data residency docs.
Architectural/technical debt indicators
• UI/API mismatch bugs (e.g., dubbing flag) and rollout naming confusion for Scribe v2 suggest rapid iteration; expect occasional documentation drift.

02Why they buy

Indications from customer/developer feedback
• Reported multilingual accuracy and performance in noisy/complex scenarios; developer anecdotes and internal/press benchmarks highlight Scribe relative to low‑latency peers.
• Enterprise privacy features: Zero Retention Mode; SOC 2/ISO; HIPAA with BAA; EU/India data residency.
Features/capabilities
• Predictive “negative latency” behavior (next‑word/punctuation prediction) to reduce effective latency; manual commit control; VAD; flexible audio encodings.
Differentiation vs competitors
• Combines accuracy, latency, and compliance requirements in one stack and integrates with ElevenLabs’ TTS/dubbing/Agents.
Where competitors have clear advantages
• Pricing (OpenAI/AssemblyAI); on‑prem deployments (AssemblyAI/Speechmatics); third‑party ASR feature breadth (varies by vendor).
Market perception vs reality
• Marketing claims of “most accurate” are supported on selected multilingual benchmarks; independent evaluations show mixed leaders depending on dataset and formatting rules. Buyers should run narrow, representative A/B tests.

• Enterprise privacy features: Zero Retention Mode; SOC 2/ISO; HIPAA with BAA; EU/India data residency.

Key Differentiator

03Field reports

Positive

“The voice quality is outstanding, natural, and expressive… the platform is fast, clean, and easy to use.” — G2 review, Oct 29, 2025.
Source: G2.
“Upgrading to Scribe significantly improved our product quality… speed tripled — a one‑hour meeting is now processed in 30–45 seconds.” — Jamie (customer story).
Source: ElevenLabs blog customer story.
“Results I get are amazingly good… already were excellent with V1.” — Reddit user discussing Scribe V2 vs V1.
Source: Reddit.

Critical/Negative

“You are charged for every single generation… cost becomes astronomical.” — Trustpilot reviewer (audiobook workflow; billing policy frustration).
Source: Trustpilot.
“Export function doesn’t seem to work… click ‘Export,’ nothing happens.” — Reddit (Speech‑to‑Text export bug report, Apr 2025).
Source: Reddit.
“The menu item disappeared… couldn’t use transcription credits… turned out to be ad‑blocker.” — Reddit (dashboard STT UI issue; later resolved by whitelisting).
Source: Reddit.
“As Scribe v2 is only real‑time… requires WebSockets.” — Reddit dev thread on integration friction.
Source: Reddit.
“API… unable to capture diarization… only Speaker 0 throughout the file.” — Reddit (Scribe v1 diarization complaint).
Source: Reddit.

Why buyers chose ElevenLabs/Scribe

Accuracy: internal/third‑party write‑ups cite lower WER vs competing low‑latency models; developers report strong multilingual performance.
Sources: ElevenLabs research post; Daily AI Primer; VentureBeat; independent blogs.
Enterprise posture: SOC 2/ISO 27001; HIPAA with BAA; Zero Retention Mode; EU/India data residency.
Sources: Trust Center; docs.

Why some left/hesitated

Pricing/billing misunderstandings or perceived expense in iterative workflows.
Sources: Trustpilot; G2 snippets; Reddit.
Occasional outages/429s in STT (load spikes; region GPU capacity).
Sources: Status page incidents.
Feature gaps/bugs (export issues, diarization inconsistencies, UI/API mismatches; docs confusion about model IDs).
Sources: Reddit threads, docs.

Specific outcomes

Jamie (meeting assistant): 3× faster pipeline; speaker errors reduced; multilingual performance.
Source: ElevenLabs blog customer story.
Early developer reports testing Realtime V2 report “amazingly good” results relative to V1 but note integration changes for WebSockets.
Source: Reddit.

Deal‑breakers mentioned

Strict on‑device or fully offline requirements (privacy or air‑gapped constraints) — cloud‑hosted Scribe is unsuitable.
- Alternative: open‑source Whisper or Speechmatics on‑prem.
  Sources: AssemblyAI (self‑hosted options), developer discussions.

Jamie (meeting assistant SaaS)

Context: Switched from a complex custom diarization/transcription pipeline to Scribe.
Results: 3× faster transcription (1h meeting → 30–45 s); complaints about speaker errors “disappeared”; multilingual coverage improved (EN/DE/ES/NL).
Implementation: “Few days” to integrate; minimal customization; reduced engineering overhead.
Challenges: Prior pipeline complexity; Scribe addressed overlapping speech and non‑verbal audio event tagging.
Source: ElevenLabs customer story.

Developer pilots (mixed)

Positive: “Amazingly good” transcriptions vs V1; language detection and accuracy strong.
Friction: Realtime requires WebSockets; incorrect parameter usage led to defaulting PCM 16 kHz; documentation/model naming confusion during rollout.
Sources: Reddit threads; realtime docs.

Operational reliability (partial outages)

Incidents: STT increased heavy‑load errors; EU data residency STT outage due to GPU availability; incidents resolved same day but highlight capacity sensitivity.
Source: ElevenLabs status pages.

Note: Independent, metric‑rich public case studies specific to Scribe v2 Realtime are limited as of Nov 26, 2025.

04Best fit

Company sizes: Startups building agents/notetakers; SMBs running multilingual support desks; mid‑market/enterprise voice operations (support/sales) where latency and compliance matter.
Industries/verticals:
• Contact centers and CX tech (real‑time agent assist).
• SaaS meeting intelligence tools.
• Media/streaming and events (live captions).
• Healthcare (with BAA and ZRM) for ambient notes/transcription (sensitive data).
Less ideal segments:
• Air‑gapped/on‑premise only environments.
• Extremely cost‑constrained batch workloads where $/min overrides accuracy/controls.
Roles/personas: Developer leads/VP Eng, Head of AI/ML, Product Managers building agents/notetakers, IT/Sec reviewing compliance (SOC 2/ISO/HIPAA, data residency).
Pain points addressed: Latency (talk‑over); multilingual accuracy; speaker diarization; enterprise privacy controls.
Qualification criteria: Need sub‑200 ms turns; multilingual; desire EU/India residency or ZRM; willing to use WebSockets/streaming.
Deal disqualifiers: Must be on‑device; strict no‑cloud; require custom on‑prem model training; require guaranteed 0% downtime (cloud dependency).
Real customer example: Jamie (meeting assistant) — faster activation/usage tied to tripled processing speed.
Source: ElevenLabs case study.

• Extremely cost‑constrained batch workloads where $/min overrides accuracy/controls.

Key Differentiator

05The pitch

What it is: Scribe is ElevenLabs’ automatic speech recognition (ASR) line. In 2025, the company launched Scribe v1 (batch STT across 99 languages) and, on November 11, 2025, introduced Scribe v2 Realtime (“Scribe2”) for low‑latency streaming ASR with ~150 ms end‑to‑end latency across 90+ languages. Product is available via API and inside ElevenLabs’ Agents platform.
Sources: elevenlabs.io blog/product pages; docs; press coverage.
Core value proposition: High‑accuracy multilingual transcription with enterprise controls. Scribe v2 Realtime targets voice agents, live captioning, call‑center assistants, and meeting assistants where latency affects user experience.
Sources: product pages, docs, launch blogs.
Target market and positioning:
• Builders of real‑time conversational agents (customer support, sales, in‑product voice UX).
• SaaS “AI notetaker” and meeting intelligence apps.
• Media/live events needing live captions and multilingual support.
• Enterprises requiring SOC 2/ISO 27001/HIPAA options, Zero Retention Mode, and EU/India data residency.
Sources: launch posts; trust center; docs.
Company background and funding: ElevenLabs (founded 2022; London/NYC) is known for TTS/voice cloning and dubbing. Series C ($180M) closed January 30, 2025 at a $3.3B valuation; in November 2025 the company announced a $6.6B employee tender valuation and reported >$200M ARR.
Sources: Reuters; TechCrunch; ElevenLabs blog; Business Insider.
Pricing model and cost considerations:
• Scribe v1 batch STT is included in subscription plans measured in “hours” (e.g., Starter $5/mo includes 12h 30m; Business $1,320/mo includes 6,000 hours; overages vary).
• Earlier public communications also cited $0.40/hour list pricing for Scribe v1.
• For real‑time (Scribe v2 Realtime), pricing is presented under ElevenLabs Agents/Realtime docs; TTS/Agents minutes appear on the broader pricing page.
• Concurrency limits for STT are higher than other ElevenLabs services (e.g., 20 on Creator, 60 on Business; Enterprise elevated).
• Note: conversational AI usage adds non‑STT costs (LLM/tooling), and ElevenLabs has adjusted voice‑agent pricing over 2025.
Sources: Scribe docs/pricing pages; blog on pricing changes; TechCrunch coverage.
Key competitors
• Deepgram (Nova‑3, Flux) — reports of real‑time accuracy and speed, per‑second billing.
• OpenAI (gpt‑4o[-mini] transcribe, Whisper) — low pricing relative to peers; ecosystem integrations.
• AssemblyAI (Universal‑Streaming) — low list price, states unlimited concurrency, self‑hosted/VPC options.
• Google Cloud Speech‑to‑Text v2, AWS Transcribe, Azure AI Speech — hyperscaler options.
• Speechmatics, Rev.ai, Gladia — enterprise/vertical plays.
Sources: competitor pricing pages and comparative articles.

Get a comprehensive analysis of ElevenLabs including market position, competitive landscape, adoption trends, and peer benchmarks.

Download full report →

Positive

“The voice quality is outstanding, natural, and expressive… the platform is fast, clean, and easy to use.” — G2 review, Oct 29, 2025.
Source: G2.
“Upgrading to Scribe significantly improved our product quality… speed tripled — a one‑hour meeting is now processed in 30–45 seconds.” — Jamie (customer story).
Source: ElevenLabs blog customer story.
“Results I get are amazingly good… already were excellent with V1.” — Reddit user discussing Scribe V2 vs V1.
Source: Reddit.

Critical/Negative

“You are charged for every single generation… cost becomes astronomical.” — Trustpilot reviewer (audiobook workflow; billing policy frustration).
Source: Trustpilot.
“Export function doesn’t seem to work… click ‘Export,’ nothing happens.” — Reddit (Speech‑to‑Text export bug report, Apr 2025).
Source: Reddit.
“The menu item disappeared… couldn’t use transcription credits… turned out to be ad‑blocker.” — Reddit (dashboard STT UI issue; later resolved by whitelisting).
Source: Reddit.
“As Scribe v2 is only real‑time… requires WebSockets.” — Reddit dev thread on integration friction.
Source: Reddit.
“API… unable to capture diarization… only Speaker 0 throughout the file.” — Reddit (Scribe v1 diarization complaint).
Source: Reddit.

Why buyers chose ElevenLabs/Scribe

Accuracy: internal/third‑party write‑ups cite lower WER vs competing low‑latency models; developers report strong multilingual performance.
Sources: ElevenLabs research post; Daily AI Primer; VentureBeat; independent blogs.
Enterprise posture: SOC 2/ISO 27001; HIPAA with BAA; Zero Retention Mode; EU/India data residency.
Sources: Trust Center; docs.

Why some left/hesitated

Pricing/billing misunderstandings or perceived expense in iterative workflows.
Sources: Trustpilot; G2 snippets; Reddit.
Occasional outages/429s in STT (load spikes; region GPU capacity).
Sources: Status page incidents.
Feature gaps/bugs (export issues, diarization inconsistencies, UI/API mismatches; docs confusion about model IDs).
Sources: Reddit threads, docs.

Specific outcomes

Jamie (meeting assistant): 3× faster pipeline; speaker errors reduced; multilingual performance.
Source: ElevenLabs blog customer story.
Early developer reports testing Realtime V2 report “amazingly good” results relative to V1 but note integration changes for WebSockets.
Source: Reddit.

Deal‑breakers mentioned

Strict on‑device or fully offline requirements (privacy or air‑gapped constraints) — cloud‑hosted Scribe is unsuitable.
- Alternative: open‑source Whisper or Speechmatics on‑prem.
  Sources: AssemblyAI (self‑hosted options), developer discussions.

Jamie (meeting assistant SaaS)

Context: Switched from a complex custom diarization/transcription pipeline to Scribe.
Results: 3× faster transcription (1h meeting → 30–45 s); complaints about speaker errors “disappeared”; multilingual coverage improved (EN/DE/ES/NL).
Implementation: “Few days” to integrate; minimal customization; reduced engineering overhead.
Challenges: Prior pipeline complexity; Scribe addressed overlapping speech and non‑verbal audio event tagging.
Source: ElevenLabs customer story.

Developer pilots (mixed)

Positive: “Amazingly good” transcriptions vs V1; language detection and accuracy strong.
Friction: Realtime requires WebSockets; incorrect parameter usage led to defaulting PCM 16 kHz; documentation/model naming confusion during rollout.
Sources: Reddit threads; realtime docs.

Operational reliability (partial outages)

Incidents: STT increased heavy‑load errors; EU data residency STT outage due to GPU availability; incidents resolved same day but highlight capacity sensitivity.
Source: ElevenLabs status pages.

Note: Independent, metric‑rich public case studies specific to Scribe v2 Realtime are limited as of Nov 26, 2025.

What it is: Scribe is ElevenLabs’ automatic speech recognition (ASR) line. In 2025, the company launched Scribe v1 (batch STT across 99 languages) and, on November 11, 2025, introduced Scribe v2 Realtime (“Scribe2”) for low‑latency streaming ASR with ~150 ms end‑to‑end latency across 90+ languages. Product is available via API and inside ElevenLabs’ Agents platform.
Sources: elevenlabs.io blog/product pages; docs; press coverage.
Core value proposition: High‑accuracy multilingual transcription with enterprise controls. Scribe v2 Realtime targets voice agents, live captioning, call‑center assistants, and meeting assistants where latency affects user experience.
Sources: product pages, docs, launch blogs.
Target market and positioning:
• Builders of real‑time conversational agents (customer support, sales, in‑product voice UX).
• SaaS “AI notetaker” and meeting intelligence apps.
• Media/live events needing live captions and multilingual support.
• Enterprises requiring SOC 2/ISO 27001/HIPAA options, Zero Retention Mode, and EU/India data residency.
Sources: launch posts; trust center; docs.
Company background and funding: ElevenLabs (founded 2022; London/NYC) is known for TTS/voice cloning and dubbing. Series C ($180M) closed January 30, 2025 at a $3.3B valuation; in November 2025 the company announced a $6.6B employee tender valuation and reported >$200M ARR.
Sources: Reuters; TechCrunch; ElevenLabs blog; Business Insider.
Pricing model and cost considerations:
• Scribe v1 batch STT is included in subscription plans measured in “hours” (e.g., Starter $5/mo includes 12h 30m; Business $1,320/mo includes 6,000 hours; overages vary).
• Earlier public communications also cited $0.40/hour list pricing for Scribe v1.
• For real‑time (Scribe v2 Realtime), pricing is presented under ElevenLabs Agents/Realtime docs; TTS/Agents minutes appear on the broader pricing page.
• Concurrency limits for STT are higher than other ElevenLabs services (e.g., 20 on Creator, 60 on Business; Enterprise elevated).
• Note: conversational AI usage adds non‑STT costs (LLM/tooling), and ElevenLabs has adjusted voice‑agent pricing over 2025.
Sources: Scribe docs/pricing pages; blog on pricing changes; TechCrunch coverage.
Key competitors
• Deepgram (Nova‑3, Flux) — reports of real‑time accuracy and speed, per‑second billing.
• OpenAI (gpt‑4o[-mini] transcribe, Whisper) — low pricing relative to peers; ecosystem integrations.
• AssemblyAI (Universal‑Streaming) — low list price, states unlimited concurrency, self‑hosted/VPC options.
• Google Cloud Speech‑to‑Text v2, AWS Transcribe, Azure AI Speech — hyperscaler options.
• Speechmatics, Rev.ai, Gladia — enterprise/vertical plays.
Sources: competitor pricing pages and comparative articles.