Hello LiveKit team! I'm currently working on a React-based project that is using LiveKit Cloud. I've been testing this project locally using Firefox and have had no issues with connecting to LiveKit Cloud. We've recently deployed our project internally and are still using LiveKit Cloud. However, while I've been accessing the project through the deployment, I'm finding it cannot consistently connect to LiveKit Cloud from Firefox. I've noticed we get repeated error messages in the console indicating it could not establish a PC connection. I also notice a similar message when I run your connection test and try connecting to the provided Websocket URL using a generated token. Interestingly, this only seems to be an issue with Firefox. When I or other users have accessed it through Edge, Chrome or Safari, it connects just fine. I would appreciate any advice or potential troubleshooting steps to help mitigate this issue. Thank you!

am I tripping or did the free plan get nerfed? Wasn't the session minutes used to be 5000 minutes on the build plan? same for the Concurrent sessions, wasn't it used to 100, but now its 5?

Hi, We’re really impressed with LiveKit and LiveKit Cloud, and we’re close to deploying virtual agents for our European customers. However, we’re currently stuck as LiveKit Cloud is only available in the us region. Could you share any timeline for when an EU region might be available?

Hi, I'm trying to deploy LiveKit agents to the LiveKit cloud and have a couple of questions: 1. For always-on agents, are we charged for the entire time they are running, or only during active conversations? 2. What is the maximum concurrency limit available on the Scale plan?

Hi everyone, sorry I am a bit new on Livekit cloud. But I built an AI assistant with Livekit that works only when I spin up the Python code to launch the server locally on my laptop. Now can someone explain me in simple steps how I can "launch" the server on the cloud using my existing Livekit Python code so I can access my AI assistant everywhere without keeping my laptop on 🙂 ? Thanks and sorry if it is a basic question.

Hello team. I recently started testing my AI Agent application using Livekit Cloud. However, I received a notification four hours ago stating “Approaching limit: 80%,” and I've just completely exhausted my 1000-minute free quota. This has me very puzzled. I used the cloud agent for less than 100 minutes today, and I had no sessions at the time the limit was exceeded. I'm not sure why this happened. Could you look into it for me? Thanks.

@great-processor-8851 I observed the same thing with my agent. I had utilized around 250 minutes, and no agent session was running, and now my 1000-minute free quota is over. Another friend of mine faced the same issue. Could someone from the team look into this?

im trying to deploy my agent to the cloud, but for some reason it keeps giving me this error "unable to get client settings: Error from intermediary with HTTP status code 401 "Unauthorized"", i followed the tutorial exactly and did everything and made sure my env.local was correct and all that, and its still not working, i even tried creating a new fresh account and it still didnt work

I have been working on a project to try and use LiveKit Cloud Agent via SIP connection. Overnight, I received an email with "Your project has exceeded its free tier limit." I figured this was an error and when logging in to check the status, I confirmed only 24 minutes used so far in October which agrees with Twilio. However, when going to looking at the Agent dashboard, it shows more than 1,000 minutes used. My telephony configuration Dispatch Rule is setup as Individual Room and in the past the Agent always drops when the caller hangs up. It looks like maybe my last test yesterday that didn't happen. Is this a bug?

Hi, I'm trying to deploy my app to google cloud run when trying to run the frontend and backend in 2 services, each one of them was built with docker. To connect between the frontend and the backend, i put in the frontend service env vars : NEXT_PUBLIC_AGENT_SERVER_URL AGENT_SERVER_URL they both got the backend service url. My code uses react and python. The backend has agent that can join a conversation if button is getting pressed. the conversations are in rooms. The problem: It's working well locally. And, when running the services in google cloud, the frontend url is working well, but when clicking on the 'add agent' button (inside a room), it gives error : Error adding agent to room: Failed to fetch TypeError: Failed to fetch. However, When i open the URL of the deployed backend , it return "OK". But, i did an endpoint /health that returns 200 status to show that the backend server is alive. But, on production, when trying to access it, it returns 404. locally its working. I don't know if the backend service deployed and running well, or maybe the services aren't connected well to each other. Or maybe there is another problem I'm missing. On the logs of the backend service there is a loop, but maybe it happens because the agent didn't need to join a call:

DEFAULT 2025-10-04T18:58:47.447493Z initializing process
DEFAULT 2025-10-04T18:58:57.747503Z 2025-10-04 18:58:57,748 - livekit.agents - INFO - killing process
DEFAULT 2025-10-04T18:58:57.747622Z killing process
DEFAULT 2025-10-04T18:58:58.647543Z 2025-10-04 18:58:57,749 - livekit.agents - ERROR - error initializing process
ERROR 2025-10-04T18:58:58.647595Z Traceback (most recent call last): File "/usr/local/lib/python3.11/asyncio/tasks.py", line 500, in wait_for return fut.result() ^^^^^^^^^^^^ File "/app/.venv/lib/python3.11/site-packages/livekit/agents/ipc/channel.py", line 47, in arecv_message return _read_message(await dplx.recv_bytes(), messages) ^^^^^^^^^^^^^^^^^^^^^^^ File "/app/.venv/lib/python3.11/site-packages/livekit/agents/utils/aio/duplex_unix.py", line 35, in recv_bytes len_bytes = await self._reader.readexactly(4) ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ File "/usr/local/lib/python3.11/asyncio/streams.py", line 750, in readexactly await self._wait_for_data('readexactly') File "/usr/local/lib/python3.11/asyncio/streams.py", line 543, in _wait_for_data await self._waiter
DEFAULT 2025-10-04T18:58:58.647598Z asyncio.exceptions.CancelledError
DEFAULT 2025-10-04T18:58:58.647602Z The above exception was the direct cause of the following exception:
ERROR 2025-10-04T18:58:58.647615Z Traceback (most recent call last): File "/app/.venv/lib/python3.11/site-packages/livekit/agents/ipc/proc_pool.py", line 167, in _proc_spawn_task await proc.initialize() File "/app/.venv/lib/python3.11/site-packages/livekit/agents/ipc/supervised_proc.py", line 169, in initialize init_res = await asyncio.wait_for( ^^^^^^^^^^^^^^^^^^^^^^^ File "/usr/local/lib/python3.11/asyncio/tasks.py", line 502, in wait_for raise exceptions.TimeoutError() from exc
DEFAULT 2025-10-04T18:58:58.647618Z TimeoutError
DEFAULT 2025-10-04T18:58:58.848165Z error initializing process

(edited)

Hello Team, A customer we are planning to work with needs all data residency in Canada. I was checking your list of Cloud Data Centers & notice Canada is not in the list. Any plans to add a data center there soon?

Hi, I’m currently based in France and using LiveKit Cloud. Could you please let me know the steps I should take to ensure my server is hosted in Europe? Thanks in advance for your help!

Hi team, We have a self-hosted LiveKit instance running on Docker in a VM with 4 CPU and 8GB RAM, and we're experiencing a critical intermittent issue where web call initialization sometimes works and sometimes fails to start at all, with no consistent pattern. The same setup works perfectly fine every time with LiveKit Cloud, but our self-hosted instance is unreliable and we cannot identify the root cause. This is affecting our production environment urgently. Could you please help us diagnose why the self-hosted instance behaves inconsistently and what configuration or resource changes we need to make it as reliable as the cloud version?

hi, We are using livekit cloud for big meetings for upto 1000 users. Now we want to live stream the meeting using HLS. Can we use livekit cloud to record the stream in realtime to S3 and then stream to users using HLS? If this is allowed what is the expected latency?

Architecture: aarch64 Platform: Linux-6.8.0-1039-aws-aarch64-with-glibc2.35 Timing Results: • Module import (MultilingualModel): 2.12s • VAD model loading: 0.24s • MultilingualModel instantiation: KO !!! I am not able to instantiate in Python in a AWS a1.large :

from livekit.plugins.turn_detector.multilingual import MultilingualModel
        model = MultilingualModel()
        duration = time.time() - start

If we are using LiveKit cloud for WebRTC transport (but not agent hosting) and a react frontend app, will we get the Krisp noise cancellation ootb?

bringing this back to the top of the discussion in case someone has an answer for me please 🙂 https://livekit-users.slack.com/archives/C048FRL1N2C/p1759674034076729

Hi team, what's the best way to set up a timer for a room and notify the user before it ends?

Hey can someone help with this, there is no way to hard refresh the session and ingress screens. Even with Auto-refresh interval set to 5 mins, it does not refresh the stats.

i am trying to deploy my python agent to livekit cloud with latest cli, i am getting following issue (venv) gauravkesharwani@Gauravs-MacBook-Pro talk2me-agent-v4 % lk agent create Using project [talk2me-phone-agent-test] Creating new agent Using secrets file [.env] Detected project type [node] Using existing Dockerfile package @livekit/agents not found in any project files

what’s the right way to structure a python project to deploy multiple agents yet share some code and have one pyproject.toml between them? Feels like I’m missing something obvious

Business question for you folks. A client I work with wants to be on the "Scale" pricing tier for their product. But, they also have three different environments (dev, qa, prod) referencing three different LK projects respectively (dev, qa, prod). If they purchase "Scale" for prod, they don't get the benefit on their dev/qa environments. But they want to get the benefits of "Scale" in all three environments so that we aren't encountering differing rate limits between environments. Now I could easily prefix room names with "dev-{room}", "qa-{room}", and "prod-{room}" and just referencing a single LK project. But then we can't effectively permission our team across environments. Some folks should have dev/qa, but not prod access to LK console. And we risk human-error code changes in dev impacting production rooms since the keys are the same. There's obviously also buying Scale in all 3 environments, but dev/qa do not share nearly the same traffic demands. Is there another, 3rd option? I imagine lots of folks have/will encounter this with multiple environments. Can we apply billing to an organization instead of an environment?

Hey team 👋 I was wondering to know if I can deploy agent in Livekit server (infra) but use a special domain name (excluding ....livekit.io)? or I should just self host it ?

Hey quick question why the cloud stopped recording at 5 am? Next screenshot is last 12 hrs, there is no data in my cloud dashboard

Hey team 👋 I have a quick question about SIP integration. Is it possible to change the domain part of the SIP URI in LiveKit (for example, using a custom domain instead of the default one)? Thanks in advance! 🙏

Hey team, my agents are no longer working after deployment to cloud via CLI, can anyone help me on this issue?

Hi team, we've noticed in our LoveKit Cloud project a whole bunch of sessions that start almost simultaneously and with only one participant (and that's just an agent), or no participants at all. And they all last 6 minutes. How can this happen? What triggers session creation if a human participant doesn't initiate it?

Any updates on timeline regarding agents' hosting within EU?

I am seeing a 700 minute call in my calls registry (active). The Twilio phone number is unreachable and the local deplopyment is getting HTTP 429´s (rate limit). Could this be a bug that is exhausting my resources? Is there a way to terminate this session? Also, the "sessions" tab seems to be showing that this session is Closed, but the calls tab shows Active. Seems like a bug (?) Any help would be greatly appreciated. Thanks!

hey team, One question, is it possible to use livekit cloud for livekit server and self host Egress? I am seeing this in the egress config:

redis:
  address: must be the same redis address used by your livekit server
  username: redis username
  password: redis password
  db: redis db

I guess livekit cloud does not expose redis credentials.