### Answer I understand the urgency of your post-call analysis latency issue. Based on the available sources, here's what I can share: According to a Discord discussion, the post-call analysis currently uses GPT-4o mini [(1) ]() . However, you mentioned you're using Claude Sonnet 4.6, which suggests you may have customized your analysis setup. For your specific use case where you need real-time data extraction before call transfer, I'd recommend considering a custom analyzer approach. You can hook into the

call_ended

or

call_analyzed

webhook events, retrieve the transcript, and send it to your preferred LLM service for faster processing . This gives you more detailed control over the prompt and potentially faster response times. The post-call analysis configuration can be customized through the agent settings. You can set custom prompts using

analysis_summary_prompt

and

analysis_user_sentiment_prompt

fields when updating your voice agent [(2) ]() . For immediate troubleshooting, I recommend: 1. Checking your specific call IDs to identify patterns in the delays 2. Reviewing your custom analysis prompts if you've set any 3. Considering implementing a webhook-based custom extraction flow for time-sensitive data that needs to be available during call transfers Since this appears to be a production performance issue that started recently, I'd suggest reaching out directly to Retell AI support with specific call IDs showing the 2+ minute delays so they can investigate the root cause. Would you like guidance on setting up a custom webhook-based extraction system for your transfer workflow?

​ To learn more, see the sources I considered: [[1] what LLM model is used to analysis transcripts for post call analysis?](https://discord.com/channels/1186424376300679209/1326892359983829075) [[2] Update Voice Agent](https://docs.retellai.com/api-references/update-agent) ​

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

Appreciate the feedback! Feel free to ask me again with additional guidance or context.