Eleven v4 API Pricing: $0.022 per 1K Characters

Eleven v4 is ElevenLabs' text-to-speech model launched 28 September 2026, with the low-latency sibling Eleven v4 Turbo. The model id is eleven_v4. Eleven v4 API pricing on the vendor page is $0.022 per 1,000 characters, struck through against a list price of $0.08 per 1,000 characters.

Eleven v4 is not on RouterPlex yet. Create a RouterPlex account for the current 60+ models. Speech models join the same prepaid key once they are listed. A chat completion will not speak. The prepaid balance, 0% markup on vendor list price, $0 top-up fee, and hard spend limit are the contract when a row exists.

Eleven v4 API pricing #

The pricing page prints 72% off until Oct 12. It does not print the year. Checked 28 September 2026, the only reading that fits the launch date is 12 October 2026. Confirm the page before you budget past that date. $0.022 / $0.08 is 27.5% of list, which is the page's 72% off wording.

Per 1M characters, the v4 promo is $22 and the list is $80. A request at the documented 10,000-character cap is $0.22 at the promo rate and $0.80 at list. Subscription plans bundle characters differently, so check the ElevenLabs pricing page for plan quotas.

Older family list prices, for contrast only: Flash and the previous Turbo family at $0.04 per 1K, Eleven v3 at $0.08 per 1K with a 5,000-character limit and 70+ languages, and v3 Conversational at $0.04. Those are different model ids.

What eleven_v4 is #

ElevenLabs calls v4 its most emotive text-to-speech model. Vendor claims from the 28 September post: instant voice clone from about 10 seconds of audio, Professional Voice Clones on v4, improved IPA, inline audio tags, and multi-speaker context. A voice recorded in one language speaking others with a native accent is also a vendor claim. Treat those as product claims, then test them on your script.

Point-event tags in the announcement include laughs, a short pause, and similar directions. The ElevenLabs docs show the exact tag syntax to use.

Expressiveness chart #

The launch chart is the share of blind pairs won on expressiveness, with ties counted as half. September 2026. The chart caption says graders judged which clip was more expressive. The blog footnote also says graders judged which sounded more natural. The percentages on the image are the expressiveness chart:

The announcement's separate line, about 75% listener preference across those four models, is the blog summary. Cite the chart when you need the per-model split. ElevenLabs also says v4 ranked number 1 on the Artificial Analysis Provider Voice Arena in September 2026.

For Google's own TTS prices, see Gemini 3.8 Flash TTS and Flash-Lite TTS. Those are also not on RouterPlex yet. Google bills tokens. ElevenLabs bills characters. Do not divide one sticker by the other and call it a winner.

Use RouterPlex today #

There is no speech route to curl. The live catalog is chat and the models already listed. This call hits claude-sonnet-5 at $2 / $10 per 1M tokens. It returns text.

curl https://api.routerplex.com/v1/chat/completions \
  -H "Authorization: Bearer $ROUTERPLEX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "claude-sonnet-5",
    "messages": [
      {"role": "user", "content": "Draft a 20-word line a narrator could read over a product demo."}
    ]
  }'

The same prepaid key is what a future Eleven row would bill against, with a hard per-key budget. See the model catalog and the OpenAI-compatible API.

On RouterPlex, once Eleven v4 is listed #

We will publish the route name, not a guessed id, the day the row exists. The price will be the vendor list price with 0% markup. The Oct 12 promo is ElevenLabs' clock. If you need eleven_v4 this week, call ElevenLabs directly. Register if you want the RouterPlex key ready for the current 60+ models and for v4 when it lands.