The latest AI news and updates of ElevenLabs — AI platform for voice synthesis, music generation, and conversational voice agents. Covering ElevenLabs's latest product updates, launches, and company news from the past 90 days.
ElevenLabs23h agoElevenLabs released a Realtime Skill for coding assistants, installed via npx skills add elevenlabs/skills. The skill teaches assistants to build agents on Eleven Realtime models, including v1 Mini for low-cost conversation and v1 Max for complex reasoning. It also provides a migration path for browser apps moving from OpenAI Realtime to ElevenLabs using the same SDK.
ElevenLabs23h agoElevenLabs recreated the voice of an ALS patient from an old home video, allowing her to speak at her daughter's wedding. The company reports its technology has now restored voices for over 11,000 people living with injury or illness. ElevenLabs has also announced a commitment to help one million people worldwide regain their ability to speak.
ElevenLabsOct 21ElevenLabs launched Eleven Realtime Translation, a live voice translation model now in enterprise alpha via ElevenAPI. The system supports 90+ languages with automatic detection, providing two-way translation where each party hears the other in their native language. The API offers 17,500+ library voices or custom clones, with an eagerness parameter to balance translation latency against output quality.
ElevenLabsOct 20ElevenLabs launched Eleven Realtime, a class of speech-to-speech models tuned end-to-end for voice agents. The v1 Mini model focuses on natural conversation, while v1 Max adds reasoning and tool-calling for complex tasks. Both models are available in ElevenAgents and via an OpenAI-compatible Realtime API, with v1 Max reaching 570ms latency and 98.8% accuracy on Big Bench Audio.
ElevenLabsOct 16ElevenLabs launched the Video to Music API, which generates music and sound effects synchronized to video cuts and action. The API includes parameters to separate music and SFX tracks and uses the video's original audio to guide the genre, BPM, and mood of the generated score. This service is now live on ElevenAPI for enterprises and developers.
ElevenLabsOct 15ElevenLabs blind-tested its new Eleven v4 voice model against competing models across eight languages. Listeners selected Eleven v4 as more expressive in every head-to-head test. The model, which also powers the low-latency Eleven v4 Turbo, uses a new architecture designed to interpret tone, pacing, and emotional context for more natural speech delivery.
ElevenLabsOct 14ElevenLabs now runs its own customer support on ElevenAgents, resolving over 22,000 tickets monthly with a 60-second initial response time. The agent handles 65% of cases end-to-end by following standard operating procedures, performing account lookups via webhook tools, and escalating to staff when necessary. The team uses ElevenAgents Architect to diagnose escalations and validate performance through simulation tests.
ElevenLabsOct 13Baroness Campbell of Surbiton delivered a speech in the UK House of Lords using an AI voice recreation built by ElevenLabs, marking a first for the UK Parliament. The voice replicates the tone, rhythm, and pitch of the Baroness, who has spinal muscular atrophy. ElevenLabs reports its technology has restored voices for over 11,000 people to date.
ElevenLabsOct 10ElevenLabs released synthetic voice detection in ElevenAgents, analyzing caller speech in the opening seconds to distinguish human from AI-generated audio. Routing rules then prioritize human callers or restrict inbound AI agents. ElevenLabs also joined the Personal Agent Protocol working group, led by Meta and Sierra, to define standards for agent-business interactions.
ElevenLabsOct 9ElevenLabs is partnering with Banner Health to deploy ElevenAgents for patient-facing call handling. The agents answer calls around the clock, booking, rescheduling, or canceling appointments directly in Banner’s electronic medical record. When human support is required, the system performs a warm transfer to a team member with the full call context, ensuring HIPAA-compliant, secure patient interactions.
ElevenLabsOct 8ElevenLabs launched Brand Kit in ElevenCreative, a workspace feature that stores logos, fonts, colors, and usage guidelines. Tagging the kit with @ in Image & Video, Flows, or the agent applies these assets automatically to new generations. Setup requires only a website URL or asset upload, ensuring consistent brand identity across all team-generated content.
ElevenLabsOct 7ElevenLabs signed its first MOU with an Indian state government to pilot voice AI, announced at its Bengaluru Summit. The company reported that over 70% of its 100M+ ElevenAgents conversations in India over the past year occurred in Hindi, Kannada, Tamil, and Telugu, highlighting the multilingual demand for voice AI in the region.
ElevenLabsOct 7ElevenLabs has launched nine text-to-speech and two speech-to-text models on OpenRouter. The integration supports models including v4, v4 Turbo, and Scribe v2 via existing OpenRouter API keys. All ElevenLabs models are available at a 50% discount on OpenRouter through October 19, 8am PT.
ElevenLabsOct 6ElevenLabs launched ElevenAgents Architect, a built-in expert for its ElevenAgents platform that builds and improves conversational agents. The tool analyzes transcripts to identify root causes, proposes simulation-validated fixes, and constructs new agents from natural language descriptions. Every change lands as a versioned draft requiring approval, and the tool integrates with Claude, ChatGPT, Cursor, and Grok Bot.
ElevenLabsOct 2ElevenLabs has integrated its Eleven v4 voice model into the ElevenReader app. The update provides expressive narration for articles, ebooks, and PDFs in 90+ languages. ElevenReader is available on iOS, Android, and the web, offering a selection of voices for imported documents.
ElevenLabsOct 2ElevenLabs has achieved FedRAMP 20x Class A certification, authorizing its ElevenAgents, Text to Speech, and Speech to Text APIs for use by US federal agencies. The certification applies when these tools operate in Zero Retention Mode with US data residency. Agencies can now verify the security evidence package directly through the company’s trust center and the FedRAMP Marketplace.
ElevenLabsSep 30ElevenLabs launched Transcript Editing for its Speech to Text API, applying natural-language instructions to transcription requests. The system returns an edited transcript alongside the original, supporting tasks like formatting, redaction, or summarization. This feature costs 30% above base transcription rates with a 10-second minimum and supports instructions up to 2,000 characters.
ElevenLabsSep 29ElevenLabs is now a launch partner for the OpenAI Marketplace, making its ElevenAgents platform available to OpenAI enterprise customers. The integration allows teams to deploy voice agents in 70+ languages and apply their existing OpenAI purchasing commitment toward the service. Interested enterprise customers can register their interest through the marketplace directory, with purchasing options becoming available soon.
ElevenLabsSep 28ElevenLabs launched Eleven v4 and Eleven v4 Turbo, its latest voice models built on a new architecture. Eleven v4 delivers emotive speech with precise control via inline tags, while Eleven v4 Turbo provides ~100 ms median inference latency for real-time agents. Both models support 90+ languages, IPA pronunciation, and voice cloning, available now across ElevenAgents, ElevenCreative, and ElevenAPI.
ElevenLabsSep 26ElevenLabs demoed a realtime scam-detection pipeline that pairs its Scribe v2 Realtime transcription with Jev’s risk-scoring engine. The system transcribes caller speech live, identifying potential fraud with ~180ms latency and a cost of $0.00043 for 11 requests. A developer guide for implementing client-side streaming with Scribe v2 Realtime is available.
ElevenLabsSep 25ElevenLabs launched an Image & Video API, providing access to 18 models including Veo 3.1 and Seedance 2.5. The API operates asynchronously via Flows, returning a generation ID immediately for webhook or polling collection. Available on Pro plans and above, the service uses the same credit system as the ElevenLabs app, with no charges for failed generations.
ElevenLabsSep 23ElevenLabs is bringing its voice, music, and video generation models to Meta’s Muse AI agent. The integration enables voiceover, soundtrack, and video generation directly within the agent via message. This partnership also extends ElevenLabs’ audio capabilities, including dubbing and character voices, across Meta’s broader ecosystem, including Instagram and Horizon.
ElevenLabsSep 21ElevenLabs launched Studio 4.0, an AI-native video editor within ElevenCreative that generates video, images, voice, music, and sound effects directly in the project. The update features a rebuilt multi-track timeline, in-editor caption editing, and a built-in Studio Agent that performs edits on request. It also adds real-time team collaboration with time-stamped comments directly on the timeline.
ElevenLabsSep 21ElevenLabs now supports agentic workflows across four integration methods: Skills, Plugins, CLI, and MCP. These interfaces provide agents with direct access to voice synthesis, music generation, and conversational tools. The Skills repository, CLI, and MCP server endpoint unify these capabilities, supporting audio task execution directly within compatible coding assistants and terminal environments.
ElevenLabsSep 17ElevenLabs reports that Cadence now uses ElevenAgents to screen vital sign alerts across 20+ US health systems. The agent reduces median patient contact time from 1 hour 48 minutes to 3.5 minutes by automating symptom screening and warm-transferring to nurses. This deployment handles 40,000 monthly alerts, resolving 60% without clinician intervention and increasing patient answer rates to 47%.
ElevenLabsSep 16ElevenLabs launched Reception, an AI receptionist platform built on ElevenAgents. The service answers incoming calls 24/7, responds to business inquiries, and manages appointment scheduling. Setup requires only a website URL, and the platform provides call summaries, transcripts, and recordings. It integrates with existing calendars to automate booking during busy periods or after business hours.
ElevenLabsSep 15ElevenLabs deployed its AI SDR, Dom, to qualify inbound sales leads using ElevenAgents Workflows. Dom achieves a 4-minute median qualification time compared to a two-day human median, with 92% accuracy. The agent generated over $1M in qualified pipeline in one month, while saving the SDR team 180 hours and handling 54% of conversations outside standard business hours.
ElevenLabsSep 14ElevenLabs integrated voice, music, image, and video generation into its Model Context Protocol. This update provides access to speech synthesis, dubbing, sound effects, and over 50 image and video models directly within assistants like Claude, ChatGPT, Cursor, Grok Bot, and Hermes. All generated assets automatically sync to the ElevenCreative workspace for further editing and refinement.
ElevenLabsSep 11ElevenLabs launched Music v2.5, featuring richer melodies, live-sounding instrumentation, and deeper arrangements. The model supports long-form compositions, mid-track genre transitions, and native-sounding multilingual vocals. It is available in ElevenMusic, ElevenCreative, and via the Music API. Every track includes commercial rights, with the Free plan offering five daily downloads and the Pro plan providing 400 lossless downloads monthly.
ElevenLabsSep 10ElevenLabs entered a multi-year strategic partnership with Universal Music Group to develop AI audio products for artists and a new fan-facing platform. Currently in development, the platform enables fans to remix, mash up, and reinterpret tracks from participating artists and songwriters. The agreement also includes joint development of AI audio tools for the creative community.
ElevenLabsSep 10ElevenLabs released CLI v1.2.0, featuring a new say command that converts text to speech and plays audio as it streams. This bypasses file writing, supporting direct piping of agent output into the terminal for immediate playback. The command defaults to the eleven_v3 model and is available now via Homebrew and npm.
ElevenLabsSep 3ElevenLabs integrates ElevenAgents into Genesys Cloud environments to power contact center interactions. The integration supports voice via the Audio Connector and text-based conversations through the Bot Connector. ElevenAgents coordinate alongside Genesys Cloud Agentic Virtual Agents, while ElevenLabs' expressive voices apply to existing Genesys agents. Configurable escalation paths route conversations to human support with full history preservation.
ElevenLabsAug 25ElevenLabs launched Composer, a section-by-section song editor for ElevenMusic. The tool enables independent editing and regeneration of specific track sections, including lyrics, tempo, and key, without affecting the rest of the composition. It supports side-by-side auditioning of multiple takes and structural rearrangement, such as swapping or doubling sections, powered by the Music v2 model.
ElevenLabsAug 25ElevenLabs released CLI v1, bringing its full API into the terminal with an agents-first design. The tool includes structured JSON outputs, schema-based contracts for coding agents, and a dry-run mode for operations. It also introduces agents-as-code for ElevenAgents — agent configurations become local files with version control, diffs, and automated deployment workflows.
ElevenLabsAug 22ElevenLabs released Eleven v3 Conversational, its most expressive realtime speech model, to general availability. The model supports 70+ languages and includes audio tags for fine-grained emotional control. It is available on ElevenAgents and ElevenAPI via Text to Dialogue WebSockets, with pricing starting at $0.05 per 1,000 characters.
ElevenLabsAug 22ElevenLabs launched an MCP connector for Claude, providing direct management of voice and chat agents. The integration includes tools to review agent performance, create and update configurations, and estimate LLM costs before deployment. OAuth authentication removes the requirement for local servers or manual API key management.
ElevenLabsAug 10ElevenLabs and Deutsche Telekom are embedding AI voice agents directly into the telecom network, launching the Magenta AI Call Assistant. This app-free tool works on any device to provide real-time translation, agentic assistance for bookings, and automatic call summaries. The service rolls out to customers in Germany later this year, with support for 50 languages planned within 12 months.
ElevenLabsAug 8ElevenLabs integrated ElevenAgents into its ElevenReader app, providing real-time voice chat for audiobook listeners. The feature uses book-specific context to answer questions without spoilers, resulting in a 24% increase in listening time. Users with five or more voice chat sessions reach a 78% book completion rate, with plot summaries representing 40% of all queries.
ElevenLabsAug 6ElevenLabs launched Dubbing v2, an AI model that translates audio and video into 90+ languages while preserving the original speaker's emotion, tone, and pacing. By conditioning on source performance rather than just transcripts, the model creates natural-sounding multilingual content. Dubbing v2 is now available for integration through the ElevenLabs API.
ElevenLabsJul 30ElevenLabs launched Character Casting for ElevenCreative Audiobooks, a feature that auto-detects characters in a manuscript and assigns distinct voices to each. The tool allows authors to preview voices on real dialogue, pre-fill pronunciations across chapters, and update every line a character speaks in one step. This streamlines the production of multi-voice, studio-quality audiobooks.
Frequently asked questions