Executive Summary (Verified October 2026 Official Vendor Documentation):
- Critical October 2026 OpenAI TTS Deprecation Notice: Per OpenAI's official model documentation (developers.openai.com/api/docs/models/tts-1), OpenAI deprecated its character-priced
tts-1($15/1M chars) andtts-1-hd($30/1M chars) models on October 1, 2026 (with removal scheduled for January 6, 2027), transitioning its audio stack toward token-priced realtime audio models.- Official ElevenLabs API Rates (
elevenlabs.io/pricing/api): ElevenLabs now publishes direct USD API rates of $0.08 per 1,000 characters for flagship expressive/multilingual models (Eleven v4, Eleven v3, Multilingual v2) and $0.04 per 1,000 characters for low-latency models (Eleven v4 Turbo at ~100ms, Flash/Turbo v2.5 at ~75ms–300ms).- Exact Per-Video Cost Mathematics:
- 1x 60-Second Vertical Short (900 characters): Costs $0.00 on ElevenLabs' 10,000-character free monthly tier; $0.036 on ElevenLabs Flash / v4 Turbo ($0.04/1k chars); or $0.072 on ElevenLabs Multilingual v2 / Eleven v4 ($0.08/1k chars).
- 1x 10-Minute Long-Form YouTube Video (9,000 characters): Costs $0.36 on ElevenLabs Flash / v4 Turbo or $0.72 on ElevenLabs Multilingual v2 / Eleven v4.
⚡ Quick Decision & Free Voice AI Credits Matrix (Verified Oct 2026)
| Voice AI Provider & Model | Official Published Latency | Official API Rate per 1,000 Chars | Exact Cost per 60s Short (900 Chars) | Official Free Tier / Link |
|---|---|---|---|---|
1. ElevenLabs (Multilingual v2 / Eleven v4) |
~250–300 ms (Long-form & emotive) | $0.080 / 1K chars (API rate) | $0.072 / Short (First 11 Shorts/mo Free) | 👉 Claim 10,000 Free ElevenLabs Chars/mo |
2. ElevenLabs (Flash v2.5 / v4 Turbo) |
~75 ms (Flash v2.5) / ~100 ms (v4 Turbo) |
$0.040 / 1K chars (50% lower rate) | $0.036 / Short | 👉 Test ElevenLabs Flash & v4 Turbo Free |
3. Cartesia (Sonic) |
< 90 ms streaming TTFA | ~$0.037 – $0.050 / 1K chars (Credit plan blended rate) | $0.033 – $0.045 / Short | Cartesia Developer Tier |
4. OpenAI (tts-1 / tts-1-hd) (Deprecated Oct 1, 2026)
|
Standard HTTP audio response |
$0.015 / 1K (tts-1) · $0.030 / 1K (tts-1-hd) |
$0.0135 – $0.027 / Short (Sunset Jan 6, 2027) | developers.openai.com/api/docs/models/tts-1 |
🛠️ Building an automated video or podcast pipeline? Connect ElevenLabs directly to n8n Cloud (14-Day Free Trial) or Make.com (1,000 Free Credits) using our binary stream configuration in Section 4.
1. Official Vendor Specifications & Pricing Comparison (Verified Oct 2026)
Every figure in the table below is sourced directly from the official vendor documentation as of October 2026—no unverified third-party claims or synthetic MOS guesses.
Table A: Official Model Specifications, Character Limits, and Published Pricing
| Specification | ElevenLabs Multilingual v2 & Eleven v4
|
ElevenLabs Flash v2.5 & Turbo v2.5 / v4 Turbo
|
Cartesia Sonic (Sonic 3.5) |
OpenAI tts-1 / tts-1-hd & gpt-4o-mini-tts
|
|---|---|---|---|---|
| Official Model Status (Oct 2026) | Active Flagship Production Models | Active Low-Latency Production Models | Active Streaming State-Space Model |
tts-1 & tts-1-hd Deprecated Oct 1, 2026 (Removal Jan 6, 2027)
|
| Published API Price per 1,000 Chars | $0.08 / 1K chars (elevenlabs.io/pricing/api) | $0.04 / 1K chars (elevenlabs.io/pricing/api) | Credit-based (~1 credit/char; ~$0.037–$0.050 / 1K chars on Scale/Pro) |
$0.015 / 1K (tts-1), $0.030 / 1K (tts-1-hd), or token-billed ($0.60/1M in + $12/1M audio out) |
| Subscription Plans (Bundled Credits) | Free (10k credits), Starter ($5/mo), Creator ($22/mo), Pro ($99/mo), Scale ($299–$330/mo) | Uses 0.5 credits per character on subscription plans (doubles character output) | Free tier, Starter, Scale, Enterprise credit tiers | Billed via OpenAI API prepaid balance |
| Official Published Model Latency | Optimized for long-form consistency |
~75 ms (Flash v2.5), ~100 ms (v4 Turbo), 250–300 ms (Turbo v2.5) |
Sub-90 ms streaming Time-to-First-Audio | Not published as a fixed ms SLA |
| Max Characters per Single Request |
10,000 chars (Multilingual v2 & v4); 3,000–5,000 chars (v3) |
40,000 chars (Flash v2.5 & Turbo v2.5) |
Streaming WebSocket / HTTP chunking | 4,096 chars per request (tts-1 / tts-1-hd) |
| Supported Languages (Official Count) |
29 languages (Multilingual v2) · 90+ languages (Eleven v4) |
32 languages (Flash/Turbo v2.5) · 90+ languages (v4 Turbo) |
42 languages (Sonic 3.5) |
Multilingual (inherits Whisper language set) |
| Self-Serve Voice Cloning | Instant Voice Cloning + Professional Voice Cloning (PVC) | Yes | Instant & Pro Cloning (~1.5x credit rate) | No public self-serve voice cloning (Preset voices only: Alloy, Echo, Fable, Onyx, Nova, Shimmer, etc.) |
| Expressive / Audio Parameter Controls |
stability, similarity_boost, style, use_speaker_boost + Audio Tags (v4) |
stability, similarity_boost, speed + Audio Tags (v4 Turbo) |
speed, emotion & pitch controls |
speed (0.25x to 4.0x) or text prompt instructions (gpt-4o-mini-tts) |
2. Exact Mathematical Cost Breakdown per Video (Shorts vs. Long-Form YouTube)
To calculate your exact monthly voiceover budget without guesswork, we use the standard broadcast narration speaking rate of 150 words per minute (WPM), which equals 900 characters per minute (including spaces and punctuation):
- 1x 60-Second YouTube Short / TikTok / Reel = 150 words = 900 characters.
- 1x 10-Minute Long-Form YouTube Video = 1,500 words = 9,000 characters.
Table B: Exact Calculated Voiceover Cost by Production Volume
| Monthly Production Volume | Total Monthly Characters | ElevenLabs Flash v2.5 / v4 Turbo ($0.04 / 1K chars) |
ElevenLabs Multilingual v2 / v4 ($0.08 / 1K chars) |
Cartesia Sonic (~$0.037–$0.050 / 1K chars) |
OpenAI tts-1-hd ($0.030 / 1K chars — Deprecated) |
|---|---|---|---|---|---|
| 1x 60s Short (Single Run) | 900 chars | $0.036 (or $0 on Free Tier) | $0.072 (or $0 on Free Tier) | $0.033 – $0.045 | $0.027 |
| 30x 60s Shorts / Month (1 Short/Day) | 27,000 chars | $1.08 / mo (or covered by $5 Starter plan) | $2.16 / mo (or covered by $5 Starter plan) | $1.00 – $1.35 / mo | $0.81 / mo |
| 100x 60s Shorts / Month (3+ Shorts/Day) | 90,000 chars | $3.60 / mo | $7.20 / mo | $3.33 – $4.50 / mo | $2.70 / mo |
| 1x 10-Min YouTube Video (Single Run) | 9,000 chars | $0.36 / video (Fits in 1 request: 40k cap) | $0.72 / video (Fits in 1 request: 10k cap) | $0.33 – $0.45 / video | $0.27 / video (Requires 3 chunks: 4,096 char cap) |
| 12x 10-Min YouTube Videos / Month (3 Videos/Week) | 108,000 chars | $4.32 / mo | $8.64 / mo (or Creator Plan $22/mo) | $4.00 – $5.40 / mo | $3.24 / mo |
Two Key Engineering Insights from Table B:
-
Single-Request Character Limits Matter in n8n: A 10-minute YouTube script (~9,000 characters) exceeds OpenAI
tts-1-hd's 4,096-character limit, forcing you to build a sentence-splitting loop and an FFmpeg audio concatenation step in your workflow. By contrast, ElevenLabsMultilingual v2andEleven v4accept up to 10,000 characters per single API call (andFlash v2.5/Turbo v2.5accept up to 40,000 characters—over 40 minutes of audio in a single HTTP request!). -
OpenAI's October 1, 2026 Deprecation Risk: Because OpenAI deprecated
tts-1andtts-1-hdon October 1, 2026 (scheduled for shutdown on January 6, 2027), hardcoding new production workflows around/v1/audio/speechwithtts-1creates immediate technical debt. Building on ElevenLabs ensures long-term API stability, instant voice cloning, and single-request 10k–40k character payloads.
3. Recommended ElevenLabs Voice IDs & Parameter Settings for Automated Channels
When calling the ElevenLabs Text-to-Speech API from n8n or Make.com, you specify the target voice right in the endpoint URL (https://api.elevenlabs.io/v1/text-to-speech/{voice_id}):
-
George (
JBFqnCBsd6RMkjVDRZzb) — Documentary, B2B SaaS Tutorials & Finance:- Recommended
voice_settings:stability: 0.45,similarity_boost: 0.80,style: 0.25,use_speaker_boost: true.
- Recommended
- Brian (
nPczCjzI2devNBz1zQrb) — Deep, High-Contrast Narration for Tech & Product Demos. - Jessica (
cgSgspJ2msm6clMCkdW9) — Upbeat Conversational Pacing for Lifestyle & E-Commerce.
4. Production n8n Implementation: Binary MP3 Synthesis to Google Drive
To generate an .mp3 voiceover inside n8n and upload it directly to Google Drive without writing temporary files to disk:
Step 1: Configure the n8n HTTP Request Node (ElevenLabs Studio Voiceover TTS)
-
Method:
POST -
URL:
https://api.elevenlabs.io/v1/text-to-speech/JBFqnCBsd6RMkjVDRZzb -
Authentication:
Header Auth-
Header Name:
xi-api-key -
Header Value:
YOUR_ELEVENLABS_API_KEY(Grab 10,000 free characters/mo at *elevenlabs.io*)
-
Header Name:
-
Send Headers:
-
Accept:audio/mpeg -
Content-Type:application/json
-
- JSON Body:
{
"text": {{ JSON.stringify($json.tts_text) }},
"model_id": "eleven_multilingual_v2",
"voice_settings": {
"stability": 0.45,
"similarity_boost": 0.80,
"style": 0.25,
"use_speaker_boost": true
}
}
-
Response Format: Set Options → Response → Response Format to
File.
Step 2: Connect the Google Drive Node (Upload File)
- Set Input Data Field Name to
data(the binary buffer returned by ElevenLabs) and set File Name to=voiceover-{{ $now.format('yyyy-MM-dd-HHmm') }}.mp3.
5. Frequently Asked Questions (FAQ)
Did OpenAI deprecate tts-1 and tts-1-hd in October 2026?
Yes. According to OpenAI's official model documentation (developers.openai.com/api/docs/models/tts-1), tts-1 and tts-1-hd were marked Deprecated on October 1, 2026, with final removal scheduled for January 6, 2027, as OpenAI transitions toward token-billed realtime audio models.
What is the official ElevenLabs API price per 1,000 characters in 2026?
Per ElevenLabs' official API pricing page (elevenlabs.io/pricing/api), Text-to-Speech costs $0.08 per 1,000 characters on flagship models (Eleven v4, Eleven v3, Multilingual v2) and $0.04 per 1,000 characters on low-latency models (Eleven v4 Turbo, Flash v2.5, Turbo v2.5).
What is the maximum character limit per API request on ElevenLabs?
Per ElevenLabs' official documentation (elevenlabs.io/text-to-speech-api), Flash v2.5 and Turbo v2.5 support up to 40,000 characters per single request, while Multilingual v2 and Eleven v4 support up to 10,000 characters per request—allowing a full 10-minute YouTube narration (~9,000 characters) to be synthesized in one API call without audio chunking.
🛠️ Want to run this voiceover pipeline without self-hosting? Start free on Make.com.
Disclosure: Some links are affiliate links — I may earn a commission at no extra cost to you.
Top comments (0)