
ElevenLabs — Eleven v4
Launched September 28, 2026, Eleven v4 is ElevenLabs' new flagship text-to-speech voice model, with Eleven v4 Turbo as a…
Launched September 28, 2026, Eleven v4 is ElevenLabs' new flagship text-to-speech voice model, with Eleven v4 Turbo as a low-latency variant for real-time use. v4 adds in-sentence direction tags — [whispers], laughs on cue, [long pause] — plus support for 90+ languages, voice cloning from roughly 10 seconds of audio, and a voice library of 17,500+ voices. It powers the ElevenCreative creator app, ElevenAgents voice agents, a REST/streaming API, and the free ElevenReader listening app.
The difference is direction: instead of flat narration, you can script whispers, laughter, pauses and emphasis inside the text itself, so audiobooks, narrated shorts and voice agents sound performed rather than read — with natural pauses and emotional inflection that older TTS systems flatten out.
Key facts: v4 Turbo is aimed at voice agents and phone UX with latency in the ~100–150ms range, making spoken back-and-forth feel near-instant. The standard v4 model trades speed for maximum expressiveness. API list pricing is ~$80 per million characters for v4 and ~$40 for v4 Turbo; the launch discount that ran through October 12 has now ended, so list prices apply.
Plans: genuine free tier — $0 for 10,000 credits/month (roughly 10 minutes of audio), with v4 included on all tiers. Paid plans: Starter $6/mo (30k credits), Creator $22/mo (121k), Pro $99/mo (600k), Scale $299/mo and Business $990/mo. Annual billing saves about two months.
Best for: creators producing voiceovers, podcast intros, narrated shorts, audiobooks and AI dubbing — and developers building voice agents who need Turbo's low latency. The platform is SOC 2/ISO 27001/GDPR compliant with HIPAA-eligible workflows, which matters for professional and client work. Who should skip it: anyone who needs hours of audio a month on a free plan.
Pros
- In-sentence direction tags — whispers, laughs on cue and long pauses — make narration sound performed, not read
- 90+ languages, a voice library of 17,500+ voices, and cloning from roughly 10 seconds of audio
- Eleven v4 Turbo for real-time voice agents at ~100–150ms latency, plus SOC 2, ISO 27001 and GDPR compliance with HIPAA-eligible workflows
Watch out for
- The standard v4 model trades speed for expressiveness, while Turbo trades some expressiveness for low latency, so pick the variant for the job
- API list pricing is per million characters, so long-form generation costs scale with volume
- The free tier covers roughly 10 minutes of audio a month, which runs out fast on regular projects
Best for
Audiobook narrators, YouTubers and podcasters who want their narration to sound directed — whispers, laughs and pauses scripted into the text — and product teams building voice agents on Turbo's near-instant replies. Who should skip it: anyone who needs hours of audio a month on a free plan.
Pricing
Genuine free tier: $0 for 10,000 credits/month (roughly 10 minutes of audio), with v4 included on all tiers. Paid plans: Starter $6/month (30k credits), Creator $22/month (121k), Pro $99/month (600k), Scale $299/month and Business $990/month. API list pricing is ~$80 per million characters for v4 and ~$40 for v4 Turbo — the launch discount ended October 12, so list prices now apply. Confirm current prices on the official pricing page.
Example
A creator scripts a 60-second horror narration with [whispers] before the reveal and [long pause] before the final line, picks a low, gravelly voice from the library, and generates audio that sounds directed rather than read — then clones the same voice for the next episode so the series stays consistent.
FAQs
What is Eleven v4 Turbo for? Real-time use such as voice agents and phone UX, with latency in the ~100-150ms range.
How many languages does Eleven v4 support? 90+ languages, with a voice library of 17,500+ voices.
Can Eleven v4 clone a voice from a short sample? Yes, voice cloning from roughly 10 seconds of audio — with the usual consent and rights requirements for real voices.
What does the free tier include? 10,000 credits/month, roughly 10 minutes of audio, with v4 included on all tiers — but no commercial licence, so publishing needs a paid plan.
Reviewed / last checked: November 10, 2026.
Compare with...
Try it on the official website — pricing and availability are confirmed there.
Visit ElevenLabs →More tools to explore
Tavus Griffin — Full-Duplex Video-to-Video AI
Announced October 1, 2026, Griffin is a full-duplex video-to-video “Human Interaction Model.” It sees and hears a caller while…
MMMeta Muse
Launched September 8, 2026, Muse is Meta's free consumer personal AI agent — built to execute tasks, not just chat.
MJMidjourney
A generative image platform for concept art, visual exploration and polished creative imagery.