I noticed that in one of my accounts the LLM model rates is increasing drastically. For example GPT 4.1 jumped from $0.07 to $0.11 per minute in 3 days in one account, while it is $0.05 in another one! This is making my call rate very high. Assistant ID with the problem: 33711217-6b1f-4ada-bdee-f3f812538b76

Hi M.Khatib, Thanks — I looked through relevant docs and past threads. Likely causes and next steps: Possible causes • Different model/region or provider selected per assistant (model pricing varies). See model enums in the API docs for providers/models: https://docs.vapi.ai/api-reference/webhooks/client-message • Fallbacks or retrying across multiple LLMs caused extra requests (seen in T-21340/T-21368). Failures can trigger higher-cost fallbacks. • Long context / large tool definitions or many turns increased token usage (seen in T-29030) so per-minute cost appears larger. • Account-specific pricing/region/credentials differences (one account using billed region or different credential causing higher cost). What I recommend you check now 1. For Assistant ID 33711217-6b1f-4ada-bdee-f3f812538b76, confirm the assistant’s configured model/provider/region in the assistant/chat config. 2. Inspect the call logs for a representative high-cost call: check model used per request, prompt/completion token counts, and any fallback invocations. (Threads show multi-LLM fallbacks and large token usage cause spikes.) 3. Verify tool definitions: ensure you’re not sending very large tool metadata every turn (can blow up tokens). 4. Compare the other account's assistant settings (model, region, credentials, timeout/retries, assistant.overrides).