ElevenLabs launched Eleven v4 and a low-latency sibling, Eleven v4 Turbo, on 28 Sep 2026. It calls v4 its "most emotive text-to-speech model yet". The post's banner carries a short deadline: "3x credits included on Creator+ until October 12". If you sell voiceovers, ad reads, narration, or voice agent demos, you've got about a week to decide whether that matters to you.
For Eleven v4 freelancers the call is fairly simple. Try v4 on two or three non-client lines before 12 Oct if you already sell voice work and you're on a plan the promo covers. Try Turbo if you build or demo voice agents. Skip it this month if your current voice setup is good enough for the clients you have. Wait if you need to know what credits cost after the promo, or you're waiting on clone consent for a client. Upgrade only after a test on your own scripts. This isn't a rewrite of our ElevenLabs for freelancers overview. It's about the new model and the next six days.
Affiliate disclosure: some tool or platform links may later be affiliate or referral links. Product facts below come from ElevenLabs' "Introducing Eleven v4" post (published 28 Sep 2026, last updated 5 Oct 2026), as of 6 Oct 2026 (SAST). Benchmark and latency figures are ElevenLabs' own claims. Re-check plans, credits, and the promo on ElevenLabs' pricing page and in your account before you rely on them. Not legal advice. Not a guarantee of clients, savings, or income.
What Eleven v4 and v4 Turbo are
Eleven v4 is a text-to-speech model built, in ElevenLabs' words, on "an entirely new architecture". The pitch is delivery. ElevenLabs says it reads tone, pacing, emotion, character, and context from the text, so the same line can come out tender, urgent, or comedic depending on what's around it.
Eleven v4 Turbo is the same technology tuned for speed. ElevenLabs aims it at live, low-latency work like voice agents, and says it was optimized together with ElevenAgents, its conversational agents platform.
Both models are, per the launch post, "available now in ElevenAgents, ElevenCreative, and via ElevenAPI". The post also says you can create a free account to start generating with either one. It doesn't spell out how v4 access or credit use differs between free and paid plans, so check that in your own account before you plan around it.
Why voice freelancers should care
Four parts of the launch matter if voice is part of what you sell.
1. You can direct the read. You can describe how a line should be delivered in plain language, and add inline audio tags. The launch post's examples are [laughs], [said angrily in French accent], [light rain], and [phone buzzing]. ElevenLabs says v4 follows them more accurately than earlier models. The full tag syntax is in the ElevenLabs developer documentation. IPA phoneme support is also "significantly improved", which helps with brand names a client wants said one way only.
2. Multi-speaker scenes sound less stitched together. ElevenLabs says v4 reads the whole scene, so speakers respond to what's just been said. If you do dialogue ads or game characters, that's the claim to test.
3. Over 90 languages, with the voice kept intact. Both models support more than 90 languages. ElevenLabs says a voice recorded in one language can now speak another with a native accent while keeping the speaker's identity, without drifting back to the source accent. For dubbing gigs, that's the headline. It's a vendor claim until you've heard it in your client's language.
4. Cloning is tighter. ElevenLabs says Instant Voice Clones can capture a voice with high fidelity from about 10 seconds of audio, with better speaker similarity than before. Professional Voice Clones (PVC) are supported on v4. Request stitching, which chains generations for long-form work, is "significantly more reliable" in Studio and the Reader App. That matters for audiobooks, where a wandering voice means a re-record.
A 10-second clone is also a reason to be careful. Before you clone a client's voice, or anyone's, read ElevenLabs' own terms and get written consent. We're not adding legal rules of our own here.
The Creator+ 3x credits window, through 12 Oct
The launch post's banner says: "3x credits included on Creator+ until October 12." That's the whole of what ElevenLabs says about it in the post we checked.
What we couldn't confirm from that post:
- exactly which plans "Creator+" covers on the current pricing page
- how the extra credits show up in your account, and whether they apply to v4, Turbo, or both
- what happens to your credit maths after 12 Oct
So don't plan a client month around the promo. Open your account, check your plan name against the pricing page, and look at your credit balance before and after a test generation. If the 3x shows up, use the window for testing, not for a big client batch you'll have to price again next month.
If you already budget API meters for other tools, treat this the same way. Our API cost guide for freelancers covers the habit: log what a real job used, then price from the log, not from a launch banner.
The benchmark claims, and how to check them yourself
ElevenLabs makes three performance claims, each with a footnote:
- Ranked #1 by Artificial Analysis on its Provider Voice Arena leaderboard (footnote dated September 2026).
- Preferred by about 75% of listeners in blind head-to-head tests against Cartesia Sonic 3.6, Inworld TTS-2, Google Gemini 3.8 Flash-Lite TTS, and Google Gemini 3.8 Flash TTS. Graders heard the same line from both and picked the more expressive and natural one. Ties counted as half.
- Turbo speed: a median inference latency of about 100ms and a median time to first speech of about 150ms. ElevenLabs measured it in September 2026 over WebSocket streaming, with identical scripts and default settings, against Cartesia Sonic 3.6, xAI TTS, Google Gemini 3.8 Flash-Lite TTS, and OpenAI GPT-4o mini TTS. Network latency was measured and removed for all systems.
That last detail matters. Your client hears network latency on a real call, so your connection and setup will add to it. We haven't measured Turbo ourselves, so we won't put a number on that.
The fair test is your own script. Run a 15- to 30-second sample through v4, the model you use now, and one rival. If Google's TTS is on your shortlist, our Gemini 3.8 Flash TTS for freelancers piece covers that side. Listen blind if you can.
Try / Skip / Wait / Upgrade sheet
Try if you already sell voiceovers, ads, or localization, and you can regenerate two or three non-client lines on v4 before 12 Oct. If your plan is covered by the Creator+ promo (check in your account), the test costs you less.
Try Turbo if you build or demo voice agents and slow responses are the complaint. ElevenLabs claims about 150ms to first speech. Test it in your own agent setup before you quote it to a client.
Skip for now if your current ElevenLabs model, or your Descript workflow, or Gemini TTS, is good enough for this month's clients. Re-doing delivered lines on a new model is churn unless a client asked for it.
Wait if you need to know what credits look like after 12 Oct before you price a retainer. Wait too if a gig depends on a Professional Voice Clone or a recognisable voice and you don't yet have consent and clearance in writing.
Upgrade only after a measured sample on your real scripts, and only through ElevenLabs' own plans page. We won't tell you a tier "pays for itself". Nobody has measured that for your work.
| Your situation | Try | Skip | Wait | Upgrade |
|---|---|---|---|---|
| Voiceover or ad-read freelancer already on ElevenLabs | Yes: 2–3 non-client lines before 12 Oct | If current output already lands with clients | If you need post-promo credit costs first | Only after a logged A/B on your scripts |
| Voice agent builder or demo maker | Yes: Turbo in your own agent stack | If latency isn't a client complaint | — | Only if agent usage repeats monthly |
| Dubbing or localization work | One sample in a language your client speaks | If no native speaker can check it | For client sign-off on accent | Only after a native listener approves |
| Audiobook or long-form narration | Test request stitching on one chapter | Mid-book: don't switch models partway | Start the next book on v4 instead | Only if a full chapter holds up |
| Cloning a client or known voice | Your own voice first | Without written consent | Yes, until consent and terms are clear | — |
| Happy in Descript or Gemini TTS | Optional side-by-side | Yes, for this month | — | No reason yet |
Get the 1-page Try v4 / Skip / Wait out the promo / Upgrade chooser
Optional printable of this free guide. Soft link until checkout. Not a "paid summary." Not a guarantee of clients, savings, or income.
The Joburg reality check
Here's how we'd run the test from a Joburg desk. It works the same anywhere.
Write a 15-second non-client script with one emotion change and one tag, like an ad read that ends on [laughs]. Generate it on v4 over a fibre line you trust, not on mobile data halfway through a client day. Download a WAV and keep that as the handoff file, not a link to a generation. Then run the same script on your current model and compare on the headphones you mix on.
If v4 wins on your script, you know before 12 Oct. If it doesn't, you've lost 15 minutes and kept your setup.
FAQ
Is Eleven v4 free?
ElevenLabs says a free account can start generating with v4 or v4 Turbo. The launch post doesn't say how free and paid plans differ for v4, so check your account and the pricing page.
What is Eleven v4 Turbo for?
Low-latency work, mainly voice agents. ElevenLabs says it was built to work with ElevenAgents and claims a median time to first speech of about 150ms, measured with network latency removed.
Does the 3x credits promo apply to me?
ElevenLabs says "3x credits included on Creator+ until October 12". Check your plan against the pricing page and watch your balance. What happens after 12 Oct isn't stated.
Is Eleven v4 better than Gemini 3.8 Flash TTS?
ElevenLabs says listeners preferred v4 about 75% of the time in blind tests against a set of rivals that included Gemini 3.8 Flash and Flash-Lite TTS. That's ElevenLabs' own test. Run your own script through both before you switch.
Sources: Introducing Eleven v4, our most emotive model (ElevenLabs blog, published 28 Sep 2026, last updated 5 Oct 2026), ElevenLabs pricing. Re-check plans, the Creator+ promo, and model availability at paste time.