### Answer I don't have specific documentation addressing the latency degradation at the 7-8 minute mark or mid-call language hallucinations with Gemini 3.0 Flash. However, there are some known issues with Gemini 3.0 Flash documented in community reports: Known Gemini 3.0 Flash Issues: - There's a reported bug with parallel function calls where Gemini 3.0 Flash fails with a

thought_signature

error, causing conversations to hang [(1) ]() - This appears to be related to Gemini's extended thinking feature and may indicate broader integration challenges For Hallucinations: AI hallucinations can stem from LLM limitations—models learn patterns without true understanding, which can produce outputs that sound believable but are factually incorrect [(2) ]() - The LLM Hallucination Rate metric in AI QA measures how often the LLM generates incorrect or fabricated information [(3) ]() Recommended Actions: 1. Try switching LLMs - Consider GPT-4.1, Claude 4.5 Sonnet, or other supported models to see if the issue is Gemini-specific [(4) ]() 2. Monitor LLM latency - Check if your LLM latency is consistently above the normal range of 500ms-900ms [(5) ]() 3. Use Conversation Flow - This provides a constrained framework that can reduce AI-generated errors [(2) ]() 4. File a support ticket with your call IDs showing the latency degradation and hallucination issues for the

team to investigate

​ To learn more, see the sources I considered: [[1] Gemini 3.0 Flash Error - "missing a 'thought_signature'" when handling parallel function call results](https://community.retellai.com/t/gemini-3-0-flash-error-missing-a-thought-signature-when-handling-parallel-function-call-results/785) [[2] Troubleshooting AI Voice Agent Issues | Retell AI](https://www.retellai.com/blog/troubleshooting-common-issues-in-voice-agent-development) [[3] Terminologies](https://docs.retellai.com/ai-qa/terminologies) [[4] Deprecating on 10/30/2025](https://docs.retellai.com/deprecation-notice/10-30) [[5] Slow Latency (3000ms, triple what it should be)](https://community.retellai.com/t/slow-latency-3000ms-triple-what-it-should-be/591) ​

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

### Answer Latency of 1220-1400ms is above the recommended target of 800ms or lower for natural conversational flow. [(1) ]() Here are strategies to reduce it further: 1. Switch to a Faster LLM LLM choice significantly impacts latency. Based on the changelog, GPT-4.1 nano has the lowest latency, followed by GPT-4.1 mini. [(2) ]() Latency ranking (lower is faster) : GPT 4.1 nano < GPT 4.1 mini < GPT 4o mini < GPT 4o < GPT 4.1 2. Check for Latency-Increasing Settings In the agent detail page under "Estimated Latency," certain settings are marked with a turtle icon indicating they increase latency. [(3) ]() Review your agent configuration and disable any non-essential features flagged this way. 3. Reduce Prompt/Knowledge Base Size Large knowledge bases in prompts can impact performance. [(4) ]() If you have extensive content in your base prompt, consider trimming it or using the Knowledge Base feature instead. 4. Optimize External Integrations If you're using custom functions or webhooks, ensure your external APIs respond quickly—slow API responses add to overall latency. 5. Recent Platform Optimizations Retell has made optimizations cutting average response times by 100-200ms and proactively avoiding latency spikes during OpenAI's peak traffic. [(2) ]() If issues persist after these optimizations, I'd recommend filing a support ticket with specific call IDs showing the high latency for the team to investigate.

​ To learn more, see the sources I considered: [[1] ](https://www.retellai.com/resources/ai-voice-agent-latency-face-off-2025) [[2] Platform Changelogs | Retell AI](https://www.retellai.com/changelog) [[3] Check estimated latency](https://docs.retellai.com/reliability/check-estimated-latency) [[4] Performance-related configurations and noticeable latency during live calls](https://community.retellai.com/t/performance-related-configurations-and-noticeable-latency-during-live-calls/747) ​

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

### Answer To create a support ticket for Retell AI, you have a few options: Option 1: Discord Support Ticket Channel You can create a ticket in the dedicated Discord channel here: [(1) ]() Option 2: Email Support You can email support@retellai.com with your issue details. [(2) ]() Option 3: Community Forum For bug reports, you can also create a ticket on the #❓・bug-report channel. [(1) ]() When submitting a support ticket for your latency issues, include: - Your call IDs showing the latency degradation - Screenshots if applicable - Details about your configuration (LLM provider, voice settings, etc.)

​ To learn more, see the sources I considered: [[1] Discord message from @elguru9007 on 2025-08-20](https://discord.com/channels/1186424376300679209/1186462578310910112/1407580307225444494) [[2] unable to log in: access denied error](https://discord.com/channels/1186424376300679209/1349125434943737928) ​

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