I would use the CLI but I can't authenticate it but idk if that would help anyways. The docs have good setup advice for claude but not codex. I got it working in vscode codex extension. Codex desktop uses the same plugins, skills, etc. so I'm surprised it doesn't work there but it does not. Screenshot of codex desktop conversation. I stored my api key in a setx variable b/c codex vscode said it was exposed otherwise. Please advise on how to setup for codex. Thank you! https://cdn.discordapp.com/attachments/1515048085599031346/1515048086110601346/Screenshot_2026-06-12_133646.png?ex=6a2d95fb&is=6a2c447b&hm=d98c180cef0c64fd6dc8760d73d9bdc1ab81781eaa7f12625a26c8376513e7ca& https://cdn.discordapp.com/attachments/1515048085599031346/1515048086576173117/Screenshot_2026-06-12_133629.png?ex=6a2d95fc&is=6a2c447c&hm=149ae0681e21cbbb6a56342cf1b7d0a9c8520ca37981fb025d011e34241f0b54& https://cdn.discordapp.com/attachments/1515048085599031346/1515048087016571031/Screenshot_2026-06-12_133649.png?ex=6a2d95fc&is=6a2c447c&hm=ac76c8d2e2b600d0087629e46c76c196821ae58ccab503c7cf9b56dbc59b6604&

Hello I have -1 concurrency counter stuck for my organization can you please reset it for me Organization ID: 9d9772f3-9eb6-4d88-a410-1b90f2fd7280

i need to change my email in my account and it does not work I dontwant to have to rebuild my stack

I'm constantly having problems with not getting latency data from VAPI. This is extremely annoying with my production calls as it makes all our dashboards completely incorrect. For example: 019eb860-1755-7001-8e3f-40c68baf6d43 This call says 0ms latency for the LLM which of course we know isn't the case. I don't understand what I can do to make VAPI give me the actual latency measurements consistently.

So I have a squad, in which there are 3 agents, One for routing, and other two, where the call would be routed after the routing agent classifies what the caller wants. The problem is that, sometimes, the routing agent doesn't transfer the call at all to the next agent, even after correctly classifying where the call needs to be routed. So there are 2 scenarios here; 1. the agent classifies where the call needs to be transferred and call is transferred sucessfully and quickly. 2. the agent classifies where the call needs to be transferred, but the call is never transferred to the agent, and the call ends after 30 seconds (the agent has 30-seconds silence timeout), but if you say something like any word, then AFTER that, the call is transferred, I want the call to be transferred each time, without fail, without the caller needing to say a filler word in between

Your server rejected

tool-calls

webhook. Error: socket hang up - but the server GOT the api request and we responded back in 6seconds...so why did this happen multiple times?

hey guys we use make.com to send http request to Vapi and after that Twilio does the call, so far all good, but the client wants to have a small short delay before the start, so I removed the first message but it took like 5-6-7 seconds at times to talk and was completely silent, so I decided to pass the first message that the client wanted client to hear, but it just shoots streight away, tried a lot of things asked the Vapi Ai agent in the website to help and he provides options for "startSpeakingPlan" { "waitSeconds" 2.0, "smartEndpointingEnabled" true } but this 2.0 even if i set it to 100 it doesnt apply any change still continue the same, did also a few other way but same result: "startSpeakingPlan" { "waitSeconds" 2.0, "smartEndpointingPlan" { "provider" "livekit", "waitFunction" "2000 / (1 + exp(-10 * (x - 0.5)))" } }, Can you please review the functionality I think there might be a something stopping the feature to work.

I got this message when trying to authenticate the CLI with

vapi login

. Your help is appreciated. https://cdn.discordapp.com/attachments/1514728603814723645/1514728603978567781/image.png?ex=6a2c6c71&is=6a2b1af1&hm=39ef2e5a52070859cad371983464d02d40040880e34287f71b2ecd870bef4c41&

Hey team, I need your help investigating a stuck call because API tooling returns 0 logs/webhook entries for it, leaving me with zero visibility. * Call ID:

019eb2f7-f441-7000-8958-187917d50530

* Status: Stuck

in-progress

for > 24h (Started:

2026-06-10T19:17:35Z

) * Type:

webCall

(Daily) | Cost:

0

### 🔍 The Core Issues: 1. No Logs Available: Querying the API for call/webhook logs for this timeframe returns absolutely nothing (

0 entries

). 2. Timeouts Ignored: The call bypassed

maxDurationSeconds: 3600

(1h),

silenceTimeoutSeconds: 40

, and

customerJoinTimeoutSeconds: 90

. ### ❓ What I need: * Can you check your backend logs for this Call ID to see why it's a "zombie" and why no logs were generated/saved?

Hi — we're evaluating Vapi for an inbound flow where calls answered on our own Twilio number (our IVR answers and parks the caller in a queue) get handed to a Vapi assistant mid-call when no human agent picks up. We already run Vapi for outbound calls in production. The documented approach for our case is POST /call with phoneCallProviderBypassEnabled: true, then applying the returned phoneCallProviderDetails.twiml to the live call leg via Twilio's call-update API — per your "Call handling with Vapi and Twilio" and "Call queue management" guides. Before we build on it, we'd like clarity on its status: Support status. phoneCallProviderBypassEnabled / phoneCallProviderDetails aren't in the public OpenAPI spec (api.vapi.ai/api-json) or in any server SDK (we checked Python 1.11.1 and TS 1.2.2), though the phone-call-provider-bypass-enabled-but-no-call-received endedReason is. A staff reply in Oct 2025 said it's "only available over the API". Is this flag fully supported and committed long-term, or at risk of being retired? Any plans to add it to the spec/SDKs? Successor? The spec's CreateCallDTO.transport accepts a TwilioTransport object ({provider: 'twilio', accountSid, callSid, callToken, forwardedFrom}), and there's a TwilioTransportMessage {twiml} via send-transport-message. Is this the intended (or future-preferred) way to attach an existing in-progress Twilio call to an assistant? If so, is there a guide? For whichever mechanism you recommend: (a) does per-call context still go through assistantId + assistantOverrides.variableValues in the same POST /call, and (b) is the returned TwiML safe to apply to an already-answered call via twilioClient.calls(sid).update({twiml}) (the queue-management guide pattern), rather than as the inbound webhook response? Gating/limits. Is either path plan-gated or subject to rate/volume limits we should know about? Thank you!

{"error":{"code":429,"message":"Resource has been exhausted (e.g. check quota).","status":"RESOURCE_EXHAUSTED"}} Callid:019eb76e-467e-7994-898e-203ffc8dcb37

Hi VAPI team, I'm building an inbound voice receptionist (Kings Fade Barbers) and have a consistent issue where the assistant goes completely silent after a tool call returns successfully. It only continues speaking once the user says something, then it responds normally. Setup: - Model: GPT-4o Mini (also tested GPT-4.1 and GPT-5.2 Instant — same issue on all, so not model-specific) - Transcriber: Deepgram Nova-3 - Voice: ElevenLabs Flash v2.5 - Custom tool hitting an n8n webhook (check_availability_kings_fade) - Tool returns successfully in 2-3 seconds What happens: 1. User gives a time to book 2. Request Start message fires correctly: 'Let me just check that for you, one moment' 3. Tool executes and returns a result successfully (confirmed in n8n execution logs) 4. Assistant goes silent — does nothing 5. User speaks 6. Assistant then responds with the tool result normally What I've already tried: - Smart Endpointing: tested both Vapi and Livekit — same issue - Silence Timeout: set to 50 seconds - Back off seconds: set to 0 - Wait seconds: set to 0.1 - Server timeout: set to 10 seconds - Temperature: set to 0 - Tested every available model — issue persists across all of them The tool response format from n8n is: { results: [ { toolCallId: 'xxx', result: 'That slot is available.' } Is there a specific turn detection setting, tool call handling configuration, or known issue with the assistant not taking its turn after a tool result is received? Any help would be greatly appreciated.

I'm having issues with variables being passed in and used correctly by the assistant. We are passing in certain keys, values, and addresses and the assistant is not using them. 019eaf7e-f4c4-744e-9fe8-1733c0e38e0a Didn't even try to use the tool 🙁 019eae73-c11a-7ddc-abec-c9209f75e049 no or little hallucinations on this call. still not perfect usage. 019eae76-c35f-7446-b075-6d631131015c said different address than variable. We expect them to use the catalog and send the id in the job tool. The assistant makes up addresses and service types. Doesn't use our ids. Other weird issues.

https://discord.com/channels/1211482211119796234/1514059902731948122 thank you

Summary: Outbound calls hitting voicemail (mailbox full) end with endedReason=customer-ended-call. We expected assistant-ended-call or a dedicated voicemail/AMD-related endedReason so we can reliably track voicemail outcomes. Org ID : 79cad7e1-c048-45c5-a126-6228d06ad29a Call IDs: - 019ead06-a410-7000-9fb6-60739ee8b44f - 019ead10-4a94-7000-8b6d-73dbbf89a284 Evidence: Both call transcripts include: "The mailbox is full and cannot accept any messages at this time. Goodbye." Immediately after that message the remote system disconnects. Observed: - endedReason: customer-ended-call (both calls) Questions: 1) Is customer-ended-call the intended endedReason when the remote voicemail platform hangs up? 2) Is there a way to get a voicemail-specific signal (endedReason or another field) for these cases? https://cdn.discordapp.com/attachments/1514230590071636068/1514230591435051089/image.png?ex=6a2a9ca2&is=6a294b22&hm=7bc12ec0ab950ecc140b3091607c14e9b412abe25bc709c9e03c909a5bd2280b&

All pages on the Dashboard is showing a error. Error: TypeError Message: Failed to fetch dynamically imported module: https://dashboard.vapi.ai/js/index-BMnozGv3.js Timestamp: 2026-06-10T07:33:43.340Z Path: /logs https://cdn.discordapp.com/attachments/1514172135848017950/1514172136263258204/image.png?ex=6a2a6631&is=6a2914b1&hm=74081697e1524b708ca47ee57354fe2e328a19756ce2bbbf10b61c0e19615983& https://cdn.discordapp.com/attachments/1514172135848017950/1514172136883884113/image.png?ex=6a2a6631&is=6a2914b1&hm=025534c78d93e06144e384433e87b22aac6642cc7f054b6d70592f6d57706e1c&

I wanted to purchase credits, but when I tried to buy them, an error message popped up saying that the card was invalid, even though I received a notification that the bank had approved the transaction and the bank had confirmed it. However, the credits were not purchased because you rejected the payment. Could you please help me with this issue?

Error: 400 Function tools with reasoning_effort are not supported for gpt-5.4-mini in /v1/chat/completions. Please use /v1/responses instead.

Do you have a solution (other than using different class of model)?

Assistant: 87be6d69-c9e9-4244-a89b-c06ab2ae8a9e Call: 019eae97-96a7-7000-a023-7c36dc144b58 On this inbound call, the first message (greeting) was logged as played — secondsFromStart 1.283, duration 8378ms — but the end-of-call cost breakdown shows ttsCharacters: 41, which is exactly the length of the second assistant line ("Hi there, what can I help you with today?"). The ~178-character greeting produced zero TTS audio. The caller heard silence, said "Hello? Hello?", and the recording contains no assistant audio for the greeting. The error field is null — no error was thrown. Voice is ElevenLabs eleven_turbo_v2_5. This is intermittent — other calls render the greeting fine. Questions: (1) Why would the first message log 8.4s "played" with 0 TTS characters and no error? (2) Does voice.fallbackPlan catch this silent-render case, or only hard provider errors? We want to make sure the greeting can never play as silence again.

Setup: - Inbound calls via Twilio SIP trunk configured as byo-phone-number in Vapi - Server-side webhook handling

assistant-request

- Responding with

destination: { type: 'sip', sipUri: 'sip:+1XXXXXXXXXX@<twilio-sip-trunk-domain>' }

to immediately forward the call without an assistant Expected behavior: Call is forwarded to the SIP URI, same as it works for Vapi-managed phone numbers. Actual behavior: Vapi responds to Twilio's SIP INVITE with 603 Decline (~2s, 0 connect duration). The call is never forwarded. Confirmed via Twilio call insight logs (

last_sip_response_num: 603

). What we ruled out: - The

destination

response is correctly received by our webhook and returned to Vapi (logged) - TWILIO_SIP_TRUNK_DOMAIN is correct and reachable (verified with a SIP capture proxy — Vapi never attempts to dial the SIP URI at all) - Not a 403/auth issue — the 603 is immediate and deliberate - Works fine on Vapi-managed phone numbers with the same response format Question: Is

destination

in

assistant-request

responses intentionally unsupported for byo-phone-number SIP trunk calls? If so, what is the recommended way to immediately forward an inbound byo-phone-number call without spinning up an assistant? The

phone-call-control

forward works but requires the call to be in-progress first.

Please see these calls: 019ead41-30c3-7000-96e3-8667da92999e 019ead3f-7d06-7000-96b9-c0fc568f78f0 The assistant's words constantly have gaps I've tried with a different LLM, like a faster LLM, a different TTS model, and a smaller prompt, and it's still the same issue

@User Zadarma Answer: Dear Juan Ayensa Barbarin, Your ticket is already being processed by the technical department. The calls shown in your examples appear to be one incoming call and one outgoing call, both processed correctly on our side. We did not receive any transfer attempt, meaning SIP REFER packets, for either of the two calls. These two calls should probably have been merged by Vapi. You can ask Vapi support: why were they not merged? You can also view and request the SIP traces in Vapi. As you mentioned earlier, “warm transfer” is an experimental Vapi feature. To learn more about how to use this experimental feature, we strongly recommend contacting the Vapi support team for any questions related to that functionality. https://cdn.discordapp.com/attachments/1513928341831549088/1513928342338797699/message.txt?ex=6a298324&is=6a2831a4&hm=2085f27361fd813eb310040398ce4a1ed73845bbdf72c395ea6ca6d179b57467&

Hi Team, I seem to be getting this error very regularly when using the VAPI portal. This happens when navigating to a new page, or openning a call log. https://cdn.discordapp.com/attachments/1513887289099419671/1513887293264629881/image.png?ex=6a295ce9&is=6a280b69&hm=beaff0932d234818ff23dcd143e2d963865d2f246e91c02d374af024504ab961&

Call ID: 019eabef-ff3c-7000-9569-553baecfe3a9 Timestamp: 2026-06-09 at 10:31 UTC Expected response: Outbound phone call connects and rings the destination number via Telnyx BYO SIP trunk. Actual response: Call goes to status "queued" but never connects. Error: call.in-progress.error-sip-outbound-call-failed-to-connect. No SIP traffic reaches Telnyx — no CDR logs appear on their side at all. Additional details: Provider: byo-sip-trunk (Telnyx FQDN, gateways 192.76.120.10 / 64.16.250.10, outbound only, options ping enabled, SIP registration realm sip.telnyx.com) All Telnyx-side configuration verified: number assigned to SIP trunk, trunk assigned to Outbound Voice Profile, destination country enabled. Direct Telnyx number import via dashboard also failed consistently with "Incorrect Telnyx Credential" errors before switching to BYO SIP trunk approach.

🚨 OUTBOUND CALLS HARD-FAILING (NO RING) — call.start.error-get-transport We cannot place ANY outbound calls in this org — fails instantly before ringing. - Happens across: Vapi numbers + Twilio + Telnyx - EndedReason: call.start.error-get-transport - Cost: $0 - Even a blank assistant test fails - Re-added credit card; still failing orgId: 286c856b-ee2a-4940-a99c-5125c5555370 Recent failing call IDs: - 019eaa13-db3a-7441-994c-256f42861611 (blank assistant test, Vapi number -> +17473898407) - 019ea817-2441-7002-96bd-b9a16a4a217a (Vapi number retry) - 019ea805-427d-7ddd-b8ab-fdd8350e3ec4 (Telnyx) - 019ea7eb-cb0c-7002-84d1-098f959adc1d (Twilio) Request: Can an engineer please check/reset outbound transport provisioning / telephony entitlement for this org? It looks like transport acquisition is failing globally.

Severity: High — all eval runs dead org-wide, blocking pre-deploy regression testing. Org ID: b0f738d5 · Account: egleyboise@gmail.com · First seen 2026-06-07, still reproducing 2026-06-08. Every Eval Run is created but never executes any turns — on BOTH the dashboard and the API: - Dashboard: run shows "Running" forever, Duration <1s, Total/Eval Turns = 0. After refresh the assistant label decays to "Unknown Assistant." - API: POST /eval/run is accepted, then GET /eval/run/{id} stays status:"queued" forever — never → running/ended; endedReason/endedMessage stay null. 100% reproduction, multiple days, every assistant. No credits consumed. Looks like the eval execution/worker layer is down for our org (queue accepts, nothing is picked up) — not a UI or config issue. Example stuck API run: 5b63803c-2f39-436a-a04d-660b43e06c72 Eval: 22425619-54ee-4bdd-a232-5cd451b82fe2 ([ETA-EVAL] ALL-01-a5) Assistant: c9880898-3c0c-4ca8-8386-f86d29e92edc Repro (dashboard): Test → Evals → pick any eval → Run → select assistant → confirm → run sits "Running" forever, 0 turns. Questions: 1) Known org-scoped issue where eval runs never leave queued/running for org b0f738d5? 2) Why does the assistant decay to "Unknown Assistant"? 3) Both dashboard + API hit the same stuck engine — what's the path to restore a working eval layer? 4) Can you share internal run-level error logs for run 5b63803c (endedReason/endedMessage are null)? Need a working eval path (or an incident ETA) to run our 16-case regression suite before shipping prompt changes. Happy to provide more run IDs. (Minor: a default user-agent got Cloudflare error 1010 from api.vapi.ai; a browser UA was needed to reach the API. Flagging in case it's edge-related.)

I'm trying to receive chat.created and session.created webhook events when a customer texts one of our Twilio phone numbers (imported into Vapi, smsEnabled: true). What I've configured: - Phone number has a server URL set - Assistant d0b95202-1b84-496f-8fff-e7df7db74d92 has serverMessages explicitly set to include chat.created, chat.deleted, session.created, session.updated, session.deleted What I observe: - Texting the number works — the assistant replies correctly via SMS - Our server receives status-update and end-of-call-report events for phone calls on the same assistant/phone number - No chat.created or session.created events arrive when the number is texted Questions: For inbound SMS on a number with smsEnabled: true, are chat.created and session.created events sent?

Hi, I'm facing an issue where I'm able to find the voice by its ID in Voice Library. However, when I'm trying to either scroll down or use the ID or even use Custom ID to find the voice in assistant configuration, it doesnot show up. This is the voice ID: bd9120b6-7761-47a6-a446-77ca49132781 - Cartesia. Please let me know how can i get this voice assigned to Assistants without any issues.

Outbound calls are failing instantly for our org across Twilio, Telnyx, and Vapi numbers. EndedReason: call.start.error-get-transport Cost: $0 (never rings) Example call IDs: 019ea817-2441-7002-96bd-b9a16a4a217a 019ea812-ae99-7002-9735-78a3e5e0096b 019ea80d-4695-7115-b39b-0c47257dca16 019ea805-427d-7ddd-b8ab-fdd8350e3ec4 019ea7eb-cb0c-7002-84d1-098f959adc1d We re-added our credit card; still failing. Can someone check org-level transport provisioning?

Hello Vapi Support Team, I hope you're doing well. I'm currently building an inbound AI receptionist using a Vapi Hosted Number and would appreciate your guidance on a specific use case. My goal is to personalize the greeting at the very beginning of an inbound call. Before the assistant speaks its first message, I would like Vapi to trigger my n8n workflow, which is connected to my CRM. The desired flow is: 1. An inbound call is received. 2. Vapi sends the caller's phone number to my n8n workflow. 3. n8n checks my CRM for a matching contact. 4. If a contact is found, n8n returns the customer's first name. 5. The assistant uses that first name in the very first greeting. Example: "Hi John, thank you for calling MD Smart Services. How can I help you today?" Currently, I have configured a Server URL, and my n8n workflow is being triggered successfully. However, the webhook appears to be receiving events such as

assistant.started

, which means the assistant has already started before my CRM lookup completes. I also tested returning values using:

json
{
  "assistantOverrides": {
    "variableValues": {
      "customer_name": "John"
    }
  }
}

but the returned variable is not available in the assistant's first message. Could you please advise: 1. Is there a supported way to perform a CRM lookup before the assistant's first message is generated? 2. For inbound calls on a Vapi Hosted Number, is there an Assistant Request event, pre-call hook, or another mechanism that allows dynamic variables to be injected before the greeting is spoken? 3. What is the recommended approach for personalizing the first greeting using CRM data retrieved from an external system such as n8n? Any documentation, examples, or recommended architecture would be greatly appreciated. Thank you for your time and support. Best regards, MD MAHIUDDIN