Unlimited Open Source Models
Get PlanLast updated · By ModelsLab Engineering
The ElevenLabs API costs $0.10 per 1,000 characters for Multilingual v2 and v3, and $0.05 for Flash and Turbo (checked September 23, 2026). That is $100 or $50 per million characters. Subscription plans run from $6 to $990 a month, and unused credits roll over for at most two months.
From the ElevenLabs API pricing page, read September 23, 2026.
| ElevenLabs model | Per 1,000 characters | Per 1M characters | ~Per minute of audio |
|---|---|---|---|
| Multilingual v2 | $0.10 | $100 | ~$0.09 |
| Eleven v3 | $0.10 | $100 | ~$0.09 |
| Flash / Turbo | $0.05 | $50 | ~$0.045 |
| v3 Conversational | $0.05 | $50 | ~$0.045 |
Per-minute figures assume about 900 characters per spoken minute (150 words), which is a typical narration pace. Your scripts may be faster or slower.
Monthly subscription tiers, as listed on elevenlabs.io/pricing and elevenlabs.io/pricing/api on September 23, 2026.
| Plan | Price / month | Credits / month (pricing page) | Characters (API page) |
|---|---|---|---|
| Entry (pay as you go) | $0 | 10k | 10,000 |
| Starter | $6 | 30k | 60,000 |
| Creator | $22 ($11 first month) | 121k | 220,000 |
| Pro | $99 | 600k | 990,000 |
| Scale | $299 | 1.8M | 2,990,000 |
| Business | $990 | 6M | 9,900,000 |
The two ElevenLabs pages list different amounts for the same plan: credits on the main pricing page, characters on the API pricing page. Confirm which model a count assumes before you size a plan.
The ElevenLabs pricing page says unused credits roll over for up to two months, up to 2x your monthly quota, so a balance can reach at most 3x the monthly quota. Credits older than that are gone. Rollover does not apply to the $0 entry plan.
In practice this means a quiet month is only banked for two more. A team that saves credits for a quarterly batch job will lose part of them, so size the plan to your steady monthly use, not to your peak.
Every generation spends characters, including the retakes you throw away. If a paragraph needs three takes to get the delivery right, it costs three times its length. Budget for your real retake rate, not for the length of the finished script.
Public reviews that mention credits or billing. Each quote links to its source; we have not verified the individual accounts. Read on 2026-09-23, when the ElevenLabs TrustScore was 3.0.
“It has bugs where simply regenerating the audio fixes the issue—forcing you to waste tokens and credits without making any changes.”
“They only supplied 5000 characters on the plan, which was useless for a narrator book-to-MP3.”
ElevenLabs direct against ModelsLab, open-weight voices and ElevenLabs voices resold on the same key.
| Option | Unit price | 1M characters (~1,111 min) |
|---|---|---|
| ElevenLabs Flash / Turbo, direct | $0.05 / 1k characters | $50 |
| ModelsLab open-weight TTS, metered | $0.001 / second of audio | ~$67 |
| ElevenLabs Multilingual v2 / v3, direct | $0.10 / 1k characters | $100 |
| ModelsLab Open Source Unlimited plan | $149 / month flat | $149 at any volume |
| eleven_multilingual_v2 on ModelsLab | $0.004 / second of audio | ~$267 |
| eleven_v3 on ModelsLab | $0.02 / second of audio | ~$1,333 |
ModelsLab prices from the live model catalogue and plans table, 2026-09-23. ModelsLab meters speech by the length of the generated audio, so the per-million figures use the same 900 characters per minute assumption (about 1,111 minutes of audio).
One POST with a voice_id and model_id. Swap model_id to move between open-weight and ElevenLabs voices.
1curl -X POST 'https://modelslab.com/api/v7/voice/text-to-speech' \2 -H 'Content-Type: application/json' \3 -d '{4 "key": "YOUR_API_KEY",5 "model_id": "eleven_multilingual_v2",6 "voice_id": "YOUR_VOICE_ID",7 "prompt": "Welcome back. Here is what changed in this release."8 }'
1import requests23response = requests.post(4 "https://modelslab.com/api/v7/voice/text-to-speech",5 json={6 "key": "YOUR_API_KEY",7 "model_id": "eleven_multilingual_v2",8 "voice_id": "YOUR_VOICE_ID",9 "prompt": "Welcome back. Here is what changed in this release.",10 },11)12print(response.json())
Comparing providers rather than prices? Read the ElevenLabs alternative guide. For the lowest-cost speech APIs across vendors, see the cheapest text to speech API comparison, and for cloning a voice from a short sample, see voice cloning API.
Providers compared on voices, latency and cloning.
Speech API prices across vendors.
Clone a voice from a short sample and use it on any text.
The ElevenLabs v3 model on a ModelsLab key.
The other direction: transcription billed per request.
TTS providers ranked by use case.
Get Expert Support in Seconds
Want to know more? You can email us anytime at support@modelslab.com
Every competitor price and quote on this page, with the date we read it.