ElevenLabs
A Tier · 8.5/10
**Nov 4: ElevenAgents now integrates natively with NiCE CXone, Genesys Cloud, Amazon Connect and Five9, and connects to any other contact center over SIP.** **SFX v3 (2026-10-29), the newest ElevenLabs sound-effects model, follows prompts that describe several sounds more closely and handles music and vocals better than SFX v2; it is in ElevenCreative (Sound Effects tab, Flows, Studio, Music Editor, mobile app) and in the API as eleven_text_to_sound_v3 on the same endpoint, though looping still needs the v2 model.** **Eleven Realtime (2026-10-20) adds two speech-to-speech agent models, v1 Mini and v1 Max, in ElevenAgents and an OpenAI Realtime-compatible API at $0.08 per minute of connected audio plus LLM tokens; Realtime Translation followed on 10/21 in alpha for API enterprise customers.** **Instant Soundtracks (2026-10-15) scores any uploaded video with original music and timed sound effects in ElevenCreative, ElevenMusic and the API, with effects on separate layers and the option to use the clip's existing audio as a style reference.** **ElevenAgents gained two new tools in the week of 2026-10-06: Architect (alpha, 10/06), a built-in expert that diagnoses agent failures from transcripts and drafts tested fixes, and synthetic voice detection (10/09), which flags AI callers in the first seconds of a call -- enterprise-only for now. ElevenCreative added Brand Kit (10/09).** **Two days after the v4 launch, ElevenLabs closed a $300M employee tender at a $22B valuation (2026-09-30), double its February Series D, with enterprise now 55% of revenue -- and the pricing page is running v4 launch promos (extra v4 credits on Creator plans and above, see the Oct 2 note).** **Eleven v4 and Eleven v4 Turbo launched 2026-09-28 -- a fourth-generation voice model on an entirely new architecture**, ranked #1 on Artificial Analysis' Provider Voice Arena and preferred by ~75% of listeners in blind head-to-head tests against Cartesia Sonic 3.6, Inworld TTS-2 and Google's Gemini 3.8 Flash and Flash-Lite TTS; v4 Turbo answers with a ~100ms median inference latency (~150ms to first speech over WebSocket) and is co-optimized with ElevenAgents. Both cover 90+ languages, follow inline audio tags like [laughs] or [phone buzzing] more accurately, clone voices with better speaker similarity, and are live now in ElevenAgents, ElevenCreative and the API (Text to Dialogue). Eleven v3 becomes the previous generation. The 2026-09-10 Universal Music Group licensing deal and CLI v1 (8/24) still stand.
Score Breakdown
The Good and the Bad
What we like
- +Voice quality is still the best available -- Eleven v4 (2026-09-28) rebuilds the model on a new architecture, ranks #1 on Artificial Analysis' Provider Voice Arena and wins ~75% of blind head-to-heads against Cartesia, Inworld and Google's 3.8 TTS pair, with a ~100ms Turbo variant for agents
- +11.ai (alpha launched June 2025, still gated through 2026) is the first serious MCP-based voice-first personal assistant -- persistent context across tasks, talk-not-type agent workflows
- +Voice cloning from just a few minutes of audio remains shockingly accurate, now with stronger consent/verification after 2025 deepfake incidents
- +~50% pricing cut in February 2026 (post-$500M raise at $11B valuation) makes the Starter and Creator tiers significantly more affordable than in 2025
What could be better
- −Eleven v4 is exposed through the Text to Dialogue API with a 10,000-character limit per request, and the launch post gave no price (the API pricing page now lists v4 at $0.08 and v4 Turbo at $0.04 per 1K characters, checked 2026-10-29); Google's 3.8 Flash TTS (9/23) ships with a public per-token rate card, which ElevenLabs' plans do not map to directly
- −Character limits mean costs still add up fast for long-form content (audiobooks, podcasts) even after the 2026 price cut
- −Free tier restricts you to personal use -- need to pay for commercial
- −11.ai is alpha-only -- not yet GA and access is gated
- −Mistral Voxtral TTS (March 2026) now offers open-source 4B-param speech for free -- the gap has narrowed for self-hosting use cases
Pricing
Free
- ✓10,000 credits/mo
- ✓3 custom voices
- ✓Eleven v4 / v4 Turbo access (2026-09-28; v3 remains as previous generation)
- ✓Personal use only
Starter
- ✓30,000 credits/mo
- ✓First month $1 was advertised until Nov 1, 2026; on 2026-11-02 the Starter card still shows it beside a countdown at zero, so treat it as ending any day
- ✓10 custom voices
- ✓Commercial license
- ✓Pricing cut ~50% in Feb 2026
Creator
- ✓121,000 credits/mo
- ✓First month $11
- ✓30 custom voices
- ✓Professional Voice Cloning
- ✓Eleven v4 expressive speech (2026-09-28) -- v3 stays available as the previous generation
Realtime API (Eleven Realtime v1 Mini / v1 Max, 2026-10-20)
- ✓v1 Mini: $0.20 input / $6.00 output per 1M tokens
- ✓v1 Max: $1.80 input / $9.00 output per 1M tokens
- ✓Burst above your concurrency limit: $0.16 per minute
- ✓OpenAI Realtime-compatible WebSocket API; also selectable in ElevenAgents
11.ai (Alpha)
- ✓MCP-based voice-first personal assistant
- ✓Persistent context across tasks
- ✓Launched June 2025 as alpha proof-of-concept; access still gated, ongoing maturation through 2026
Enterprise (IBM watsonx)
- ✓Agentic voice for enterprise via IBM partnership (March 25 2026)
- ✓Regulated-industry voice cloning
- ✓Volume pricing
Known Issues
- ELEVENAGENTS GOES NATIVE IN FOUR CONTACT-CENTER PLATFORMS (2026-11-04, vendor-primary): 'ElevenAgents now integrates natively with four leading contact center platforms: NiCE CXone, Genesys Cloud, Amazon Connect, and Five9. ... For any other platform, ElevenAgents connects over Session Initiation Protocol (SIP).' The pitch is that voice agents run 'inside the call flows they already trust' -- existing telephony, routing, compliance recording and CRM stay in place, and the contact center passes call context to the agent at the start of the call. ElevenLabs cites 'more than 17,000 voices' in 'over 90 languages'. No pricing change was announced in the post.Source: ElevenLabs (elevenlabs.io/blog/elevenagents-contact-center-integrations, JSON-LD datePublished 2026-11-04; RSS pubDate Wed, 04 Nov 2026) -- fetched 2026-11-05 · 2026-11-04
- SFX V3 (2026-10-29 in ElevenCreative and ElevenAPI; developer post 2026-11-02, vendor-primary): 'SFX v3, the newest ElevenLabs sound effects model ... handles prompts that describe multiple sounds more reliably, generates music and vocals more convincingly, and produces cleaner, more natural audio than SFX v2 in benchmark testing.' In ElevenCreative it is in the Sound Effects tab, Flows, Studio, the Music Editor and the mobile app. **API:** 'set model_id to eleven_text_to_sound_v3 on the POST /v1/sound-generation endpoint'; the model picks the length unless duration_seconds is set, prompt_influence defaults to 0.3, and **'Looping currently only works with eleven_text_to_sound_v2'**. ElevenLabs itself calls it 'an incremental update' that is less likely to clip on gongs, chimes and kettle whistles, and suggests A/B testing v2 against v3 on the same endpoint. No price change was announced and no benchmark numbers were published.Source: ElevenLabs blog (elevenlabs.io/blog/sound-effects-v3, JSON-LD datePublished 2026-10-29; elevenlabs.io/blog/sfx-v3-for-developers, datePublished 2026-11-02), fetched 2026-11-02 · 2026-10-29
- ELEVEN REALTIME: TWO SPEECH-TO-SPEECH AGENT MODELS WITH A PER-MINUTE PRICE, AND REALTIME TRANSLATION IN ALPHA (2026-10-20 and 2026-10-21, vendor-primary): 'Eleven Realtime is a new class of Speech to Speech models' in two sizes, v1 Mini and v1 Max; 'Each model is a complete voice agent', both support 90+ languages and any of the 17,000+ voices, and the API 'is compatible with OpenAI's Realtime API' (WebSocket at wss://api.elevenlabs.io/v1/convai/realtime) or you pick them in the ElevenAgents model dropdown. The underlying LLM is not named. ElevenLabs cites Artificial Analysis results: Big Bench Audio 98.8% (Max) and 75.9% (Mini); time to first speech 0.57 s and 0.62 s; tool-use completion 87.1% and 60.3%. **Price (API pricing page):** $0.08 per minute of connected audio plus tokens, v1 Mini $0.20 / $6.00 and v1 Max $1.80 / $9.00 per 1M input / output tokens, burst $0.16 per minute above your concurrency limit. **Realtime Translation (10/21):** 'available in alpha for ElevenAPI enterprise customers', built on Eleven v4 Turbo, 90+ languages with automatic detection, Deutsche Telekom as a launch customer, no public price. A 10/28 post explains Cue, the alignment model already behind the Scribe family (not a new product).Source: ElevenLabs blog: Introducing Eleven Realtime (elevenlabs.io/blog/realtime, JSON-LD 2026-10-20) + Talk to anyone, in any language, in realtime (elevenlabs.io/blog/realtime-translation, JSON-LD 2026-10-21) + elevenlabs.io/pricing/api, fetched 2026-10-29 · 2026-10-20
- PLAN PRICES CORRECTED (checked 2026-10-29, vendor-primary): the pricing page FAQ reads 'Free $0 (10,000 credits); Starter $6 (30,000 credits); Creator $22 (121,000 credits, $11 for the first month); Pro $99 (600,000 credits); Scale $299 (1,800,000 credits, 3 seats)'. Our page had Starter at $5 with 30,000 characters and Creator at 100,000 characters. Starter currently shows a $1 first month until Nov 1, 2026.Source: elevenlabs.io/pricing (title 'ElevenLabs Pricing for Creators & Businesses of All Sizes'), fetched 2026-10-29 · 2026-10-29
- INSTANT SOUNDTRACKS -- MUSIC PLUS TIMED SOUND EFFECTS FOR UPLOADED VIDEO (2026-10-15, vendor-primary): 'Generate a soundtrack and synchronized sound effects for any video you upload in ElevenCreative.' Upload a video and 'the model scores it with original music, then analyzes the footage for sound effect cues, and generates sound effects aligned to those timestamps' (ElevenLabs' example: 'a whoosh timed to the logo at the end of an ad'). Effects land as separate layers in the ElevenCreative Music Video Editor so they can be adjusted or swapped without regenerating; the score can shift mood across scenes and can use the video's existing audio (genre, crescendos, BPM, instrumentation) as a reference; timing is tuned for short-form cuts. **Availability:** 'available immediately' in ElevenCreative's music tab, Creative Engine for ads, ElevenMusic, and the **ElevenCreative Studio API** via new parameters add_sfx, separate_tracks (returns a ZIP of music and SFX stems) and use_audio_as_reference; **up to 10 videos per request, 200 MB and 10 minutes combined**. No separate price was stated -- generation draws on existing plan credits. Other ElevenLabs posts this window (SSML, deepfake detection, transcription and VAD explainers, a UK Parliament AI-voiced speech) are guides or customer stories, not product changes.Source: ElevenLabs (elevenlabs.io/blog/introducing-instant-soundtracks, JSON-LD datePublished 2026-10-15; blog RSS pubDate Thu 15 Oct 2026) -- fetched 2026-10-15 via curl · 2026-10-15
- ELEVENAGENTS ARCHITECT (ALPHA), SYNTHETIC VOICE DETECTION, BRAND KIT -- AND THE v4 3x-CREDIT PROMO ENDED 10/12 (2026-10-06 to 2026-10-12, vendor-primary): **(1) ElevenAgents Architect (10/06)** -- 'our new built-in expert that helps your team launch and improve agents, just by talking or typing'. It knows every ElevenAgents configuration option (knowledge base, prompts, workflows, voice settings, procedures, guardrails, tools, simulations), takes a signal such as a failing test, an escalation spike or a Spotlight finding, analyzes transcripts, proposes a change and writes simulations to validate it. Every change lands as a versioned draft that needs human approval; teams can set standards it must follow. Reachable inside ElevenAgents and from Claude, Claude Code, ChatGPT, Cursor and Grok Bot. **'Available now in Alpha'**; no price stated. **(2) Synthetic voice detection (10/09)** -- analyzes a caller's speech in the first few seconds and returns human or AI-generated, so rules can route machine callers to a scoped agent, queue them behind people, or block sensitive actions (bank-detail changes, password resets) requested by a synthetic voice. ElevenLabs notes all its own audio is watermarked but detection does not rely on a watermark. **Available today only to enterprise customers supported by its Forward Deployed Engineering team; a broader enterprise rollout 'later this month' as a configurable option.** ElevenLabs also joined the **Personal Agent Protocol** working group (led by Meta and Sierra) as a design partner. **(3) Brand Kit (10/09)** -- colours, fonts, logos with usage rules, reference images and brand rules stored once per workspace and tagged with @ in Image & Video, Flows and the ElevenCreative agent; multiple kits for agencies; voice and sound kits 'coming'. **(4) Promo end:** the site banner still read 'With 3x credits included on Creator+ until October 12' on 10/12, so the Eleven v4 launch credit promo recorded on this page has now run its course; standard credit costs apply from 10/13.Source: ElevenLabs blog: elevenagents-architect (JSON-LD datePublished 2026-10-06; 'available now in Alpha'), synthetic-voice-detection-and-personal-agent-protocol and introducing-brand-kit (rss.xml pubDate Fri, 09 Oct 2026); site banner text on the Architect page -- all fetched 2026-10-12 via curl with browser UA · 2026-10-09
- $22B VALUATION ON A $300M EMPLOYEE TENDER, ENTERPRISE AT 55% OF REVENUE, AND TWO DIFFERENT V4 CREDIT PROMOS ON THE SAME PRICING PAGE (2026-09-30 and 2026-10-02, vendor-primary): **Tender (9/30, updated 10/01):** 'We recently closed a $300 million employee tender offer that values ElevenLabs at $22 billion, double our valuation at our Series D in February', led by Wellington and T. Rowe Price with BDT & MSD participating. ElevenLabs says **enterprise is 'now accounting for 55% of our revenue'** as ElevenAgents adoption grows, and that its technology is in daily use at 'five of the world's ten largest tech companies, five of the ten largest insurers, and four of the ten largest telecoms'. No product or price change in the post; it leans on v4 and v4 Turbo 'already leading independent benchmarks'. **Promos observed on elevenlabs.io/pricing on 10/02 -- two of them, with different terms:** the site banner reads **'Introducing Eleven v4 ... With 3x credits included on Creator+ until October 12'**, while the pricing page body reads **'Try v4 with 2x credits for two weeks -- On Creator plans and above, up to 2x your monthly TTS credits used on v4 won't count against your balance. Available in the web and mobile apps only.'** Both are vendor copy on the same page; the multiplier (2x vs 3x) and the window (two weeks vs a fixed October 12 end) do not agree, so treat the banner's 3x-until-10/12 as the more specific claim and the 2x text as possibly stale. Either way the promo is **Creator and above, v4 only, web and mobile apps only (not the API)**, and v4 remains capped at 10,000 characters per request. Plan prices on the page were unchanged.Source: ElevenLabs (elevenlabs.io/blog/tender-22bn, JSON-LD datePublished 2026-09-30, on-page 'Published Sep 30, 2026 / Last updated Oct 1, 2026') + elevenlabs.io/pricing (banner and 'Try v4 with 2x credits for two weeks' section) -- fetched 2026-10-02 via curl with browser UA · 2026-10-02
- ELEVEN V4 AND ELEVEN V4 TURBO -- NEW ARCHITECTURE, #1 ON ARTIFICIAL ANALYSIS, ~100MS TURBO LATENCY, 90+ LANGUAGES, LIVE ACROSS AGENTS, CREATIVE AND THE API (2026-09-28, vendor-primary): 'Today we're launching Eleven v4, our most emotive text-to-speech model yet, and its low-latency variant, Eleven v4 Turbo.' Claims: 'Ranked #1 by Artificial Analysis' (footnote: Provider Voice Arena Leaderboard, Sept 2026) and 'preferred by ~75% of listeners in blind head-to-head tests over competing models' (footnote: against **Cartesia Sonic 3.6, Inworld TTS-2, Google Gemini 3.8 Flash-Lite TTS, and Google Gemini 3.8 Flash TTS**, September 2026 -- i.e. against the Google models launched five days earlier). 'Built on an entirely new architecture', v4 'was designed to interpret tone, pacing, emotion, character, and context'; multi-speaker dialogue where 'speakers respond to the context of the conversation'; inline tags such as [laughs], [said angrily in French accent], [light rain], [phone buzzing] are followed 'more accurately than prior models'; IPA phoneme support 'significantly improved'; voice cloning 'with significantly better speaker similarity'; request stitching for long-form 'significantly more reliable' in Studio and the Reader app. **Eleven v4 Turbo**: 'median inference latency of ~100ms' (docs; excluding application and network latency) and 'median time to first speech of ~150ms' over WebSocket streaming, 'built to work with ElevenLabs' conversational agents platform, ElevenAgents' -- research and engineering 'optimized Eleven v4 Turbo and ElevenAgents together as one system'. Both models support **more than 90 languages** (v3 was 70+). **Docs (elevenlabs.io/docs/models):** model IDs `eleven_v4` ('via the Text to Dialogue API', **10,000 character limit**, natural multi-speaker dialogue) and `eleven_v4_turbo` ('via the Text to Dialogue websocket'); `eleven_v3` is now 'our previous generation speech synthesis model'. **Availability:** 'Both models are available now in ElevenAgents, ElevenCreative, and via ElevenAPI'; a free account can generate with either. **Not stated:** any change to per-character or credit pricing for v4 vs v3, so the plan rows keep their character quotas; and the AA leaderboard rank is a dated snapshot (see miss #19 -- re-check before repeating it). Context: this lands the same month as Google's 3.8 TTS pair (9/23, with a published per-token rate card) and Gemini Notebook/Vids distribution, so the voice-model race is now explicitly head-to-head on both quality claims and price.Source: ElevenLabs (elevenlabs.io/blog/eleven-v4, JSON-LD datePublished 2026-09-28T12:00:00Z; elevenlabs.io/docs/models -- Eleven v4 / v4 Turbo sections and model-ID table) -- fetched 2026-09-28 via curl with browser UA · 2026-09-28
- ELEVENLABS AND UNIVERSAL MUSIC GROUP -- FIRST MAJOR-LABEL DEAL, A LICENSED FAN-REMIX PLATFORM IS COMING, AND IT PUTS ELEVENLABS IN DIRECT COMPETITION WITH SUNO v6 (2026-09-10, vendor-primary): ElevenLabs announced 'a **multi-year licensing agreement and strategic collaboration with Universal Music Group**', described as 'our first agreement with a major label, encompassing licensing and product development.' **What is actually being built:** 'a new AI-powered music creation platform built on licensed music and artist participation', now in development, letting fans 'co-create with music from participating artists and songwriters, including through **remixes and mashups, new track interpretations, and personalized vocal experiences**', plus jointly developed audio products for artists and songwriters. UMG's Lucian Grainge is quoted; CEO Mati Staniszewski frames it as artists and songwriters being 'fairly compensated'. **SCOPE, PER ELEVENLABS:** the new platform 'will be distinct and offered separately from our existing music products' -- the **Music API** and **ElevenMusic** are unaffected and remain the current offering. **No launch date, no pricing, no artist roster named.** **WHY IT MATTERS:** on 9/09 Suno launched v6 with Warner, BMG and Believe and promised opt-in artist experiences; a day later UMG -- the label suing Suno -- picked ElevenLabs as its licensed-AI partner. The licensed-fan-remix category now has two label-backed entrants a day apart, and UMG is on the ElevenLabs side of it. ElevenLabs has not disclosed whether the UMG catalogue will be used to train new models or only to license outputs; the post says 'built on licensed music', which is a product statement, not a training-data one.Source: ElevenLabs (elevenlabs.io/blog/umg, JSON-LD datePublished 2026-09-10) -- fetched 2026-09-14 via curl · 2026-09-10
- STALENESS CATCH-UP, JUNE-AUGUST 2026 (this page was last reviewed 2026-07-18; everything below is from ElevenLabs' own blog, dated by JSON-LD, and none of it was on the page): **ElevenLabs CLI v1 (2026-08-24)** -- 'brings the ElevenLabs API directly into your terminal', designed 'agents first' with structured JSON output, a `--dry-run` mode, and **agents-as-code for ElevenAgents**: pull every agent in a workspace into local config files, diff, and push to production by hand or from a coding agent. Every endpoint in the OpenAPI spec is a subcommand. Install via Homebrew (`elevenlabs/tap/elevenlabs`), Scoop, or a curl installer. **Procedures in ElevenAgents (2026-06-30)**, **Ads Engine in ElevenCreative (2026-06-22)** -- localise ads across 50+ languages, **Flows Agent in ElevenCreative (2026-06-04)**, plus the previously recorded Music v2 and Dubbing v2 (May). **Corporate:** **$500M ARR** crossed with new investors (2026-05-05); **$22M earned by voice creators**, doubling in six months (2026-05-22); expansions in Canada (7/07), California (173 jobs, 6/22), Australia/NZ, plus Poland and UK government partnerships (June); a new **CRO (Ashley Kramer, 9/02)** and **CFO (Ethan Tandowsky, 9/08)** -- both ex-executive hires that usually precede an IPO-readiness push, though ElevenLabs has said nothing about listing. **No pricing changes and no model deprecations found in the window** -- the models page state recorded 7/18 (scribe_v1 deprecated, v1 TTS removed) stands.Source: ElevenLabs blog (elevenlabs.io/blog/elevenlabs-cli-v1 datePublished 2026-08-24; /procedures 2026-06-30; /introducing-ads-engine-in-elevencreative 2026-06-22; /introducing-flows-agent 2026-06-04; /500m-arr-and-new-investors 2026-05-05; /22-million-earned-by-voice-creators-on-elevenlabs 2026-05-22; /canada 2026-07-07; /expanding-in-california 2026-06-22; /cro 2026-09-02; /cfo 2026-09-08) -- all fetched 2026-09-14 via curl · 2026-08-24
- DEPRECATION STATUS SETTLED (verified on vendor docs 2026-07-18): **scribe_v1 is now formally listed in the 'Deprecated models' table** -- 'First generation speech recognition (outclassed by v2 models)', replacement suggestion `scribe_v2` -- with NO removal date published (the ambiguous removal wording from early July is gone; it remains deprecated-but-available). Current STT flagships are Scribe v2 and Scribe v2 Realtime. Also now marked deprecated/legacy on the same models page: **eleven_turbo_v2_5, eleven_turbo_v2, and music_v1**. The monolingual_v1 + multilingual_v1 removals executed 7/9-7/10 stand. If you still call any v1 or turbo_v2-era model id, plan migrations now rather than waiting for a removal date to be announcedSource: ElevenLabs docs: Models (elevenlabs.io/docs/overview/models, scraped 2026-07-18) · 2026-07-18
- V1 MODEL REMOVAL EXECUTED (2026-07-09; confirmed via docs delisting 7/10): **eleven_monolingual_v1 and eleven_multilingual_v1 are now delisted entirely from the vendor models page** -- the July 9 removal went through as scheduled, and pinned API calls to those TTS v1 model IDs should be treated as failing. **scribe_v1 remains listed as deprecated** ('outclassed by v2 models') with the removal-date wording now gone -- if you're still pinned to scribe_v1, migrate to scribe_v2 immediately rather than betting on the ambiguity. No past-tense changelog entry was posted for the removal; the delisting is the confirmation. These are model deprecations, not voice-library removals; cloned/library voices are unaffected. Separately (7/6 changelog): the Agents 'Simulate conversation' endpoints are deprecated in favor of newer test endpoints, and the `disable_interruptions` param was replaced by `interruption_mode`Source: ElevenLabs docs (elevenlabs.io/docs/overview/models -- checked 2026-07-09 and 2026-07-10), ElevenLabs changelog (2026-06-08 removal announcement) · 2026-07-09
- PRODUCT (2026-05-26 + 2026-05-28): **Eleven Music v2** (5/26) -- genre switching mid-track, coherent fast rap, embedded sound effects in generated tracks (see the elevenmusic page for detail). **Dubbing v2** (5/28) -- next-gen dubbing pipeline on the main platform. Also: UK Government voice-AI partnership announced 6/8. No pricing changes attached to either releaseSource: ElevenLabs blog (elevenlabs.io/blog), TechCrunch (2026-05-27) · 2026-05-28
- Platform continues to face deepfake-abuse pressure -- voice cloning requires verified identity for new accounts as of 2026Source: The Verge · 2026-01
- 11.ai alpha has intermittent latency issues on longer agentic chains -- still maturingSource: Product Hunt 11.ai threads · 2026-03
Best for
Content creators who need the highest-quality voiceovers, audiobook producers, developers building voice-enabled apps, and enterprises using IBM watsonx wanting premium agentic voice. 11.ai alpha users who want voice-first AI assistants.
Not for
Users who only need occasional text-to-speech (browser TTS is free), or open-source purists (Mistral Voxtral fills that niche now).
Our Verdict
ElevenLabs remained the clear voice-quality leader through 2026 and, with Eleven v4 and v4 Turbo (2026-09-28), re-founded that lead on a new architecture rather than another v3 tune: #1 on Artificial Analysis' Provider Voice Arena, ~75% blind preference over Cartesia Sonic 3.6, Inworld TTS-2 and Google's brand-new Gemini 3.8 Flash and Flash-Lite TTS, 90+ languages, and a ~100ms Turbo path co-designed with ElevenAgents. The February 2026 $500M raise at $11B, the ~50% price cut, the 8/24 CLI and the 9/10 Universal Music Group deal round out a company that now competes on catalog rights and agent latency as much as on raw voice quality. The pressure is real, though: Google's 3.8 TTS pair arrived five days earlier with voice design from a prompt, 30-second replication and a published $9-per-1M-audio-token rate card, and Mistral's Voxtral covers the open-source end. For content creators and agent builders ElevenLabs is still the default; the gap is now measured in points of preference, not generations.
Sources
- ElevenLabs: ElevenAgents integrates natively with leading contact center platforms (2026-11-04) (accessed 2026-11-05)
- ElevenLabs: Introducing SFX v3 in ElevenCreative and ElevenAPI (2026-10-29) (accessed 2026-11-02)
- ElevenLabs: SFX v3 API -- eleven_text_to_sound_v3 on /v1/sound-generation (2026-11-02) (accessed 2026-11-02)
- ElevenLabs pricing (Starter first-month $1 card still live 2026-11-02) (accessed 2026-11-02)
- ElevenLabs: Introducing Eleven Realtime (2026-10-20) (accessed 2026-10-29)
- ElevenLabs: Realtime Translation alpha (2026-10-21) (accessed 2026-10-29)
- ElevenLabs API pricing: Realtime v1 Mini and v1 Max, v4 and v4 Turbo per-character rates (accessed 2026-10-29)
- ElevenLabs pricing: Starter $6 (30,000 credits), Creator $22 (121,000 credits) (accessed 2026-10-29)
- ElevenLabs: Introducing Instant Soundtracks (2026-10-15) -- music + synced SFX for uploaded video, Studio API add_sfx / separate_tracks (accessed 2026-10-15)
- ElevenLabs: ElevenAgents Architect -- an expert partner to improve agents (2026-10-06, alpha) (accessed 2026-10-12)
- ElevenLabs: Defining how AI agents interact with your business -- synthetic voice detection, Personal Agent Protocol (2026-10-09) (accessed 2026-10-12)
- ElevenLabs: Introducing Brand Kit (2026-10-09) (accessed 2026-10-12)
- ElevenLabs blog RSS feed (exact pubDates) (accessed 2026-10-12)
- ElevenLabs: valuation increases to $22 billion fueled by enterprise demand for conversational agents (2026-09-30) -- $300M employee tender, enterprise 55% of revenue (accessed 2026-10-02)
- ElevenLabs pricing page (fetched 2026-10-02) -- v4 credit promos on Creator and above: banner '3x until October 12', body '2x for two weeks' (accessed 2026-10-02)
- ElevenLabs: Introducing Eleven v4, our fastest and most emotive voice model (2026-09-28) -- v4 + v4 Turbo, AA #1, ~75% preference, 90+ languages, availability (accessed 2026-09-28)
- ElevenLabs docs: Models -- eleven_v4 (Text to Dialogue API, 10,000-character limit) and eleven_v4_turbo (~100ms); v3 marked previous generation (fetched 2026-09-28) (accessed 2026-09-28)
- ElevenLabs: ElevenLabs and Universal Music Group enter strategic agreement -- first major-label deal, licensed fan remix platform in development (2026-09-10) (accessed 2026-09-14)
- ElevenLabs: ElevenLabs CLI v1 -- agents as code, entire API in terminal (2026-08-24) (accessed 2026-09-14)
- ElevenLabs: Introducing Procedures in ElevenAgents (2026-06-30) (accessed 2026-09-14)
- ElevenLabs: Introducing Ads Engine in ElevenCreative -- 50+ languages (2026-06-22) (accessed 2026-09-14)
- ElevenLabs: ElevenLabs crosses $500M ARR and welcomes new investors (2026-05-05) (accessed 2026-09-14)
- ElevenLabs docs: Models (scribe_v1/turbo_v2/music_v1 deprecated; v1 removals executed) (accessed 2026-07-18)
- ElevenLabs official site (accessed 2026-04-16)
- Voice.ai: ElevenLabs debuts 11.ai (accessed 2026-04-16)
- IBM Newsroom: ElevenLabs + IBM watsonx (accessed 2026-04-16)
- G2 Reviews (accessed 2026-04-16)
- Hands-on testing (including 11.ai alpha) (accessed 2026-04-16)
Explore more ElevenLabs rankings
Deeper leaderboards, benchmarks, task-specific tier lists, and status/pricing pages for ElevenLabs.
The Tier List Tuesday
Weekly newsletter: tier movers, new entrants, and the VS of the week. Built from our daily AI-tool sweeps. No spam, unsubscribe anytime.
Alternatives to ElevenLabs
Murf AI
Text-to-speech that actually sounds like a real person read your script -- not a robot trying its best
Descript
Edit audio and video by editing text -- the 'Google Docs of media editing' actually lives up to the hype
Speechify
Text-to-speech reader that turns articles, docs, and PDFs into natural-sounding audio
Microsoft MAI-Voice-2
Microsoft's in-house expressive TTS line. **MAI-Voice-2.1 and MAI-Voice-2.1-Flash launched 2026-10-01: 23 languages and 26 locales with one voice that keeps the same speaker and a native accent across languages, voice cloning from a few seconds of reference audio with consent guardrails, priced at $22 per 1M characters for 2.1 and $15 per 1M characters for Flash, which generates 45 seconds of audio at 150 ms end-to-end latency and is pitched at voice agents alongside MAI-Transcribe-2-Streaming.** These supersede **MAI-Voice-2 (Build, 2026-06-02)**: 15 languages, emotion tags, zero-shot cloning from a 5-60 s clip, preferred over MAI-Voice-1 72% of the time. Microsoft also shipped a MAI-Voice-2-Flash and MAI-Image-2.5-Pro in July that this page had not recorded -- see the Oct 2 note.
Grok Speech (STT + TTS APIs)
**Grok Voice 3.0, xAI's new speech-to-speech model, shipped 2026-10-14 at the same $0.08 per minute as 2.0, which xAI's rate card now flags as on its way out; grok-voice-latest switched over automatically, and xAI cites a #1 score of 86.9 on Artificial Analysis' Speech-to-Speech Index.** xAI's standalone voice APIs. **Grok Voice Transcribe 2.0 shipped 2026-09-18** -- xAI calls it twice as accurate as 1.0 at the same $0.10/hr batch and $0.20/hr streaming, ranks it first for accuracy among 32 streaming models on Artificial Analysis, and has already made it the default STT model; **1.0 is being retired in the coming weeks, so pin grok-voice-transcribe-1.0 if you need it**. **PRICE CORRECTION: text-to-speech is $15.00 per 1M characters on xAI's live rate card (verified 2026-09-21), not the $4.20 this page carried since April.** Speech-to-speech agent $0.08/min, 26 flagship voices, ~1-minute voice cloning, no-code Voice Agent Builder
GPT-Live (ChatGPT Voice)
**GPT-Live-1.1 and a smaller GPT-Live-1.1 Quip began rolling out in ChatGPT Voice on 2026-10-15, with more natural accents, shorter spoken cues while it searches, and new voices for French, German, Hindi, Japanese, Korean and Mexican Spanish plus an English voice called Bubbie.** OpenAI's full-duplex voice models (launched in ChatGPT 2026-07-08) -- listens and speaks at the same time, backchannels naturally, and delegates hard questions to a backend model while keeping the conversation going. **GPT-Live-1 reached the API on 2026-09-10 at $0.05 per minute** for the voice layer, with telephony support, keyword biasing, delegation to any backend model (GPT-6 Astra, Luna, or a third-party model) and a wider voice set. In ChatGPT: GPT-Live-1 for paid tiers, GPT-Live-1 mini for Free. **Five days after the API launch, Google priced Gemini 3.8 Live at $0.005/min audio in and $0.018/min audio out (2026-09-15)** -- a fraction of GPT-Live-1's $0.05/min voice layer, though each stack bills backend or text tokens on top
Cohere Transcribe
**Nov 2026: an Arabic fine-tune (cohere-transcribe-arabic-07-2026) shipped on 2026-07-07 with Arabic-English code-switching, Model Vault now publishes per-GPU hourly prices, and Cohere signed its Aleph Alpha combination agreement on 2026-09-16 (closing pending).** Cohere's first audio model -- launched 2026-03-26 under Apache 2.0, 2B parameters, #1 on the Hugging Face Open ASR Leaderboard at launch (5.42 avg WER as of 2026-03-26), 14 enterprise-critical languages. Free trial API (rate-limited, non-commercial); Model Vault for production