https://vapi.ai/ logo
Join Discord
Powered by
New Gladia integration ETA
# support
p
v
I couldn't find any updates or ETA on when the new Gladia transcriber will be integrated with Vapi, so I'm escalating this to a human teammate who should be able to help you out. Just for context, Gladia is a transcriber (speech-to-text) provider available on Vapi, and their platform offers real-time transcription and audio intelligence features. As soon as I hear back from the team, I'll let you know!
p
Hey, Could you please schedule a brief call at your convenience so we can discuss this matter in detail? Kindly use the following link to book a suitable time: .
p
Just booked for june 3nd (earliest date). if you happen to have anything earlier ping me 🙂
j
Happy to join that call if that is not too intruding - also waiting and pushing for this for ages
@Shubham Bajaj still confident it will be out this week ? 🤞🏼
p
Hello Jeebs, The issue has been listed within our team's watch list and has been postponed until next week. We apologize for any inconvenience this may cause. Please note that this is a priority item for next week.
j
Right, thanks for the transparency.
v
Update on Gladia Solaria Integration: The latest information from the Vapi team indicates that the integration of the new Gladia transcriber model (Solaria) has been postponed until next week, but it is now a priority on their team's watch list. There is no confirmed ETA or release date yet. Please stay tuned for further updates from the team as they work to complete this integration. Source: - [Vapi Changelog and Documentation](https://docs.vapi.ai/providers/transcriber/gladia)
Update on Gladia (Solaria) Integration: The latest information indicates that the integration of the new Gladia transcriber model (Solaria) with Vapi has been postponed until next week, but it is now a priority for the team. There is no confirmed ETA yet, but you will be updated as soon as there is progress. Thank you for your patience and understanding. Source: - [Changelog and team communication](https://docs.vapi.ai/changelog/2025/5/23)
v
Message marked as helpful by @pierreh! 🎉
j
Hello @Shubham Bajaj, quick msg to confirm this will be shipped this week ? 😇
v
Update on Gladia Solaria Integration: The latest official information indicates that the integration of the Gladia Solaria transcriber with Vapi has been postponed until next week and is now a team priority. However, there is no confirmed release date or guarantee it will ship this week. Please stay tuned for further updates from the Vapi team as they work to complete this integration. Source: [Vapi Changelog and Documentation](https://docs.vapi.ai/providers/transcriber/gladia) Source: - [Vapi Changelog and Documentation](https://docs.vapi.ai/providers/transcriber/gladia)
r
i also hope this gets fixed as soon as possible 🙂 @User
p
Thank you for your patience as we work to resolve this issue.
j
Kyle, @Shubham Bajaj, the new Gladia transcriber model hasn't been integrated yet, correct ? Couldn't see anything in the changelog or in the API definitions 😢
p
same, still waiting 😢
p
p
@Shubham Bajaj
s
Hey @pierreh ! To help track down this issue, could you try with a call and share: - The call ID This would really help us figure out what went wrong!
r
@User They probably aren't supporting real streaming for the transcription provider's WebSocket API, even though Gladia supports it.
Instead, they likely send short chunks of audio (like 4 to 5 sec) via standard HTTP post requests. This matches my own tests. I measured the time and from the moment someone speaks it takes about 4 seconds before the conversation updates in the context window.
This "chunked" approach introduces unnecessary latency. They probably wanted to ship it quickly without doing a full real time integration
To build a full duplex, low latency pipeline usually takes weeks to develop and test properly (managing partial results, reconnect logic, error handling). Chunking can be done in days, so this was likely a trade off to release it faster
s
@pierreh a reminder to share the call_id, this ticket is open with us and Gladia team.
p
Got an error when using gladia
p
Hey @pierreh, thanks for sharing the call_id. Checking with Gladia Team.
p
any news?
j
@Shubham Bajaj same issue here : I have a huge difference between latency in Gladia's playground and the one served via VAPI. Example call id : 131c4731-2c5f-4ed9-8801-092745fb0a42 Also, could you double check that custom vocabulary was implemented ? It seems to me that the "transcriptionHint" attribute (see VAPI API schema below) doesn't work with the solaria model - probably from the previous models. Here is the API for the new model : https://docs.gladia.io/api-reference/v2/live/init with attribute realtime (https://docs.gladia.io/api-reference/v2/live/init#body-realtime-processing-custom-vocabulary-config). https://cdn.discordapp.com/attachments/1372191439735226408/1384119458573975572/image.png?ex=68514532&is=684ff3b2&hm=d025a01d73321b616253affef643c54a75266772bb93b7be8dcc94e028727294&
p
The issue was related to the maximum number of concurrent sessions. We are in talk to increase it. Meanwhile, you can use your own key and it will solve the issue.
v
Hi vapi team, I tried 
solaria-1
with my own key but call fail as soon as they start with this error: > An error occurred in this call: pipeline-error-gladia-transcriber-failed Example Call ID: fed3f950-8dc1-4a2e-bb6f-d3246763c540 This is my configuration:
Copy code
"transcriber": {
    "model": "solaria-1",
    "prosody": false,
    "language": "de",
    "provider": "gladia",
    "languages": [
      "de"
    ],
    "audioEnhancer": false,
    "languageBehaviour": "manual",
    "confidenceThreshold": 0.4
  },
I could see in my Gladia dashboard, that a "‎Real-time Transcription" was initiated but was never complete and was hanging with status: "Processing".
p
any news on this task?
p
It is fixed, can you try again?
j
Heyy, seems like there's issue with Gladia transcriber again? It seems unstable with only half the call going smoothly cries It is currently the only transcriber model where we can specify a set of languages. Can you support a couple of alternatives please if Gladia is not working out well?
p
Can you share the call_id?
j
e293d554-5c20-4574-b124-d04b2810a0c3
the transcript also only have user message, and not assistant
p
Seems like one weird call . If you hear the audio, the last half was empty. Are you able to constantly reproduce it?
j
Taking this opportunity to push again for the basic features of Gladia which have not been integrated. The main one is the custom vocabulary (similar to keywords of deepgram): https://docs.gladia.io/chapters/live-stt/features#custom-vocabulary
p
@jaibyjeebee Please submit a feature request to the roadmap if you would like to see this on our platform -->
j
Just added it : https://roadmap.vapi.ai/feature-requests/p/add-custom-vocabulary-feature-to-gladia-transcriber I had pushed for it on the original Gladia Solaria feature request... I now have put a second one 😓
p
p
Hey pierreh, Kyle's actually out on an offsite right now, but he'll be able to dive into this for you on Monday, and help figure out the best approach for your setup. Regards, Shubham Bajaj Solutions Engineer
p
thanks, looking forward!
p
Copy code
🔴 12:18:10:539 Failed to initialize Gladia V2 WebSocket: { "message": { "message": "Request failed with status code 402", "data": { "statusCode": 402, "timestamp": "2025-07-11T12:18:10.443Z", "path": "/v2/live", "message": "You used all your 10h of free audio transcription. Please visit app.gladia.io to upgrade your plan !"
We will have to look into this more since the call was using our API Key and received this error in the logs. In the meantime for now, we would suggest using another transcriber provider while we investigate this.
p
thank you!
p
Marketing this ticket as solved ✅
PreviousNext
Powered by