### Answer To end a call using the inbound webhook, you can return a dynamic variable that indicates the call should hang up, then configure your agent to check for this variable [(1) ]() . The recommended approach is to return a dynamic variable like

SHOULD_HANG_UP: "true"

in your webhook response, and then configure your agent to check for this variable and end the call accordingly . Here's how your webhook response would look [(2) ]() :

json
{
  "call_inbound": {
    "dynamic_variables": {
      "SHOULD_HANG_UP": "true"
    }
  }
}

However, note that if your inbound webhook returns an error status (non-2xx) , the call may still connect without the dynamic variables rather than immediately ending [(1) ]() . The system is designed to gracefully degrade if the webhook fails, allowing the agent to continue rather than failing completely . For Conversation Flow agents, you'll need to create a condition in your first node to check this variable and route to an end call action if it's set to true .

​ To learn more, see the sources I considered: [[1] The call connects despite the fact that Inbound Call Webhook returns an error](https://discord.com/channels/1186424376300679209/1343523617744486511) [[2] Inbound webhook](https://docs.retellai.com/features/inbound-call-webhook) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

### Answer Yes, you can reject inbound calls before they reach the agent using the inbound webhook [(1) ]() . To selectively reject inbound calls : 1. Unset the Inbound Call Agent in the phone number setting 2. Enable the Inbound Webhook 3. In your webhook logic, check the

from_number

in the request body 4. **To reject the call**: respond with a 200 status code without the

call_inbound.override_agent_id

field 5. **To accept the call**: respond with a 200 status code containing the

call_inbound.override_agent_id

field When you don't include the

override_agent_id

in your response and there's no default inbound agent set on the number, the call will be disconnected before connecting to any agent .

​ To learn more, see the sources I considered: [[1] Inbound webhook](https://docs.retellai.com/features/inbound-call-webhook) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

Thanks for the feedback, glad I was helpful.

### Answer To set up warm call transfer in Retell AI, follow these steps: For Single/Multi-Prompt Agents: 1. Add Transfer Call Tool - Click "+ Add" in the tools section and select "Transfer Call" from the dropdown menu [(1) ]() 2. Setup Transfer Target - Set the transfer number to either a number in e.164 format, a SIP URI (e.g.

sip:user@retellai.com

) , or a dynamic variable 3. Update the Prompt - Include transfer conditions in your agent's prompt, such as: "If the user is angry or frustrated, use the transfer_call tool to transfer the conversation to a human agent" 4. Configure Transfer Type - Choose "Warm transfer" instead of "Cold transfer" 5. Configure Warm Transfer Settings : - On-hold music - Choose audio played while caller is on hold (default is ringtone) - Enable human detection - Agent checks if a human is present after transfer target answers - Auto-greet - Agent immediately says "Hello" when transfer target picks up - Agent detection timeout - Maximum time to wait to determine if transfer target is human (default 30 seconds) - *Whisper message (optional) * - Private message spoken to transfer target before connecting caller - *Three-way message (optional) * - Message spoken to both transfer target and caller once connected For Conversation Flow Agents: Use the Call Transfer Node and configure similar settings, choosing "warm transfer" as the transfer type [(2) ]() Advanced Warm Transfer: Retell AI also offers an advanced warm transfer feature where a transfer AI has a two-way conversation with the transfer target before bridging the original caller [(2) ]() [(3) ](https:/

/www.youtube.com/watch?v=pRZcA6P_4tk)

​ To learn more, see the sources I considered: [[1] Transfer call](https://docs.retellai.com/build/single-multi-prompt/transfer-call) [[2] Call Transfer Node](https://docs.retellai.com/build/conversation-flow/call-transfer-node) [[3] Advanced Warm transfer Announcement]() ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

### Answer The key difference between warm and cold transfers in Retell AI is how context and information are shared during the handoff [(1) ]() [(2) ]() [(3) ]() . Cold Transfer: - The call is simply forwarded to the destination number without sharing context [(4) ]() - The receiving agent has no information about what the caller discussed with the AI [(2) ]() [(3) ]() - The caller must repeat their problem and information to the new agent [(1) ]() [(2) ]() - The conversation effectively restarts from the beginning [(1) ]() [(3) ]() - Uses SIP INVITE or SIP REFER methods [(4) ]() Warm Transfer: - Context is preserved and shared with the receiving agent before connecting the caller [(1) ]() [(2) ]() - The AI can deliver a private whisper message to the receiving agent with caller context (e.g., "This is Alex Rodriguez canceling Monday's app

ointment due to a doctor visit") [(1) ]() - Optionally includes a three-way handoff message where the caller hears the AI briefing the agent - Features human detection to confirm a person answered before completing the transfer [(1) ]() [(4) ]() - Includes on-hold music while the caller waits - The receiving agent joins with full situational awareness, allowing the conversation to continue seamlessly [(1) ]() [(2) ]() In Retell AI, warm transfers ensure customers don't have to repeat themselves, reduce frustration, and create a more professional handoff experience .

​ To learn more, see the sources I considered: [[1] Warm Transfer vs Cold Transfer - The AI Advantage in Call Centers | Retell AI](https://www.retellai.com/blog/effortless-handoffs-with-retell-ais-warm-transfer-feature) [[2] How AI Voice Agents Are Perfecting the Warm Transfer: Bridging the Gap Between AI and Human Agents | Retell AI](https://www.retellai.com/blog/how-ai-voice-agents-are-perfecting-the-warm-transfer) [[3] Can Voicebots Escalate Calls to Human Agents? Understanding AI Voice Escalation Workflows | Retell AI](https://www.retellai.com/blog/can-voicebots-escalate-calls-to-human-agents) [[4] Call Transfer Node](https://docs.retellai.com/build/conversation-flow/call-transfer-node) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

### Answer The default and recommended LLM for Retell AI voice agents is gpt-4.1 [(1) ]() [(2) ]() [(3) ]() [(4) ]() . When creating or updating a Retell LLM, if you don't specify a model, it defaults to gpt-4.1 [(1) ]() [(2) ]() . Available LLM Options: For text-based models, you can choose from : -

gpt-4.1

(default/recommended) -

gpt-4.1-mini

-

gpt-4.1-nano

-

gpt-5

-

gpt-5-mini

-

gpt-5-nano

-

gpt-5.1

-

gpt-5.2

-

gpt-5.4

-

gpt-5.4-mini

-

gpt-5.4-nano

-

claude-4.5-sonnet

-

claude-4.6-sonnet

-

claude-4.5-haiku

-

gemini-2.5-flash

-

gemini-2.5-flash-lite

-

gemini-3.0-flash

For speech-to-speech models (set separately from text models) : -

gpt-realtime-1.5

-

gpt-realtime

-

gpt-realtime-mini

Important Note: You can only set either a text model (

model

) or a speech-to-speech model (

s2s_model

) , not both . For tool calling scenarios, a lower

model_temperature

value is recommended to ensure more deterministic responses .

​ To learn more, see the sources I considered: [[1] Update Retell LLM](https://docs.retellai.com/api-references/update-retell-llm) [[2] Create Retell LLM](https://docs.retellai.com/api-references/create-retell-llm) [[3] Get Retell LLM](https://docs.retellai.com/api-references/get-retell-llm) [[4] List Retell LLMs](https://docs.retellai.com/api-references/list-retell-llms) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

### Answer I don't have specific information about what the community is using or recommending for LLMs in the available sources. The official documentation states that gpt-4.1 is the default and recommended LLM [(1) ]() [(2) ]() [(3) ]() , but I don't have data on community preferences or real-world usage patterns. For community insights and recommendations, you might want to: - Check out discussions in the Retell AI community forums - Join the Retell AI Discord server where you're currently asking questions - Look at case studies or user testimonials on the Retell AI website The platform changelog [(4) ]() shows that various models have been added over time (GPT-5 family, Claude models, Gemini models) , but it doesn't provide usage statistics or community recommendations. If you're looking for guidance on which model to choose, consider: - gpt-4.1 for the default recommended option [(1) ]() [(2) ]() [(3) ]() - Lower temperature values for tool calling scenarios - Your specific use case requirements (latency, cost, intelligence level)

​ To learn more, see the sources I considered: [[1] Update Retell LLM](https://docs.retellai.com/api-references/update-retell-llm) [[2] List Retell LLMs](https://docs.retellai.com/api-references/list-retell-llms) [[3] Create Retell LLM](https://docs.retellai.com/api-references/create-retell-llm) [[4] Platform Changelogs | Retell AI](https://www.retellai.com/changelog) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.

### Answer I cannot directly access or search the Discord community in real-time. However, I can share what's available in the sources I have access to: From the Discord community snapshot [(1) ]() : One user mentioned they're experiencing issues with the LLM playground when testing with GPT-4o [(2) ]() . The error occurs when updating prompts with dynamic variables and then going to test . This suggests that GPT-4o is being used by at least some community members, though they're encountering technical issues. From Retell AI's blog comparing LLM options [(3) ]() : The blog discusses various LLM options for voice AI agents, comparing: - **OpenAI's GPT-4o**: Achieved a score of 88.7 on the MMLU benchmark and sets new records in audio speech recognition and translation - **Anthropic's Claude**: Excels in understanding nuanced and complex instructions, including humor and subtle context The blog notes that GPT-4o is ideal for customer service bots, virtual assistants, and interactive conversational tools requiring rapid and coherent responses . Cost comparison from the blog : - **GPT-4o**: $0.00250 per 1,000 input tokens; $0.01000 per 1,000 output tokens - **GPT-4o Mini**: $0.000150 per 1,000 input tokens; $0.000600 per 1,000 output tokens - **Claude 3.5 Sonnet**: $0.06 per minute For real-time community recommendations and current usage patterns, I'd suggest asking directly in the Discord channels or checking recent discussion threads where users share their experiences with different models.

​ To learn more, see the sources I considered: [[1] Discord message from @lucawa0009 on 2026-01-02](https://discord.com/channels/1186424376300679209/1186462578310910112/1456530986878898403) [[2] Problem with LLM playground](https://discord.com/channels/1186424376300679209/1348019311180120266) [[3] Choosing the Best LLM for Your AI Voice Agents | Retell AI](https://www.retellai.com/blog/choosing-the-best-llm-for-your-voice-ai-agents) ​

Feel free to tag @AI Evy with additional questions. I also appreciate feedback, it helps me improve.