V3 from ElevenLabs isn't working

Support Help
1

The voice function v3 from ElevenLabs isn’t working, it’s charging me for being activated but it’s not working.

2

Hey @dremlabs Could you please share your organisation ID and any related call ID?

Thank You

3

Workspace ID: org_AFGqXQU92JTNLArD

Call ID: call_23b9d23663576a2fe7d181b0ae6

4

@dremlabs Checking with the team.

5

Hey @dremlabs We found the issue on our side.

We currently have a temporary reroute in place: ElevenLabs’ V3 model is being served as eleven_flash_v2_5 due to an access issue with our ElevenLabs V3 key. That reroute went live on July 1, so your calls have been reaching the caller using Flash v2.5, not V3.

The bug is that our billing pipeline captured the configured model (V3) instead of the served model (Flash v2.5), so you were charged the V3 rate for Flash output. Call call_23b9d23663576a2fe7d181b0ae6 was billed at V3 pricing for ElevenLabs TTS when it should have been billed at Flash v2.5 pricing.

Next steps on our end:

  1. We’re fixing the billing capture so the rerouted rate is charged while the V3 access issue is resolved.
  2. We’ll credit back the V3-vs-Flash difference on affected calls under agent_de08729bc50c4055dd2c4ded62 (and any other agent on your workspace configured for eleven_v3) since July 1.

Until V3 access is restored, we can’t guarantee true V3 output for real-time calls — V3 is also not recommended for real-time phone agents (it’s designed for expressive/generative use, not low-latency streaming). If you’d like consistent expressive output right now, eleven_turbo_v2_5 or eleven_multilingual_v2 on your same Raquel clone are the recommended stable options.

Thank You

6

Am I going to be able to use Eleven Labs V3 again?

7

Hey @dremlabs yes, we’re actively working on restoring ElevenLabs V3 access. Until it’s back, agents configured with eleven_v3 will continue to be served via eleven_flash_v2_5.

Thank You