Eleven v4 lists at $80 per million characters: is Turbo at $40 enough?

On ElevenLabs' API pricing page, Eleven v4 and v3 list at $0.08 per 1,000 characters, which is $80 per million, and the Turbo and Flash models list at $0.04, or $40 per million. Eleven v4 and v4 Turbo launched on September 28, and a 72% launch discount, quoted at $0.022 for v4 and $0.011 for Turbo, runs until October 12. Turbo is enough when you need speed and a lower bill more than the top model's expression.

The ElevenLabs launch post puts Turbo at about 150 ms median time to first speech, which matters for live agents but not for a finished voiceover. For batch narration, the question is quality per dollar, and only a listening test on your own script can answer it.

What one million characters costs

Eleven figures are from the ElevenLabs pricing page and launch post. The Sume row is computed from the repository: Cartesia's list price times Sume's 1.25 margin, rounded up to the cent per job.

• After October 12 the launch discount ends, so re-run your budget on the list rows.
• The ElevenLabs blog also advertises 3x credits on Creator and higher plans until October 12; that applies to plans, not to the API rows above.

Test Turbo against the full model in four steps

• Pick one script that includes names, numbers and an emotional line.
• Render it with v4 and with Turbo, using the same voice.
• Have two people rate the takes blind for clarity and feel.
• If the scores match, use Turbo and keep the saved difference.

import os, requests

r = requests.post(
    "https://api.sume.com/v1/tts-router/generate",
    headers={
        "Authorization": f"Bearer {os.environ['SUME_API_KEY']}",
        "Idempotency-Key": "tts-demo-001",
    },
    json={
        "model": "sonic-3.6",
        "transcript": "Welcome back. Today we compare three prices.",
        "voice": {"id": os.environ["SUME_VOICE_ID"]},
        "timestamps": {"words": True},
    },
    timeout=30,
)
r.raise_for_status()
print(r.json())

Reading the Sume row

Sume's price is flat per character on Sonic models. At 1,000,000 characters it sits between Turbo's list price and v4's list price, and above the discounted launch rows. It is a different engine, so a price comparison is only half the story; listen to both.

A worked example

Take a 12,000-character audiobook sample. At the list rates that is $0.96 on v4 and $0.48 on Turbo, a gap of 48 cents per sample. On Sume it is about 57 cents (12 x $0.0475), rounded up to a whole cent on the job. The gap is small for one sample and large across a library of a thousand chapters, which is when a blind test pays for itself.

Be careful about promotional rates. A launch discount that ends on October 12 is a reason to test now, but a poor reason to commit a production budget. Build the sheet on list prices and treat any discount as a bonus that you re-check each week.

What Sume does not do

Sume does not offer Eleven models. Sume TTS covers Cartesia Sonic only. The router has no Eleven, OpenAI, Gemini, MAI or Inworld engines, no streaming TTS, and no routing presets. Jobs are asynchronous, and any audio over 1,200 seconds fails with tts_duration_exceeded. The ElevenLabs numbers here are list prices from the page I read; plan, volume and contract terms can change what you pay.

Related

See the price ranking per 1,000 characters and how to pin your model id.