Repeated issues in calls. Reproducible in multiple ways. Occurs for Gemini Flash 3. Example Logs from a particular call: 2026-01-05 13:10:25.837 call_184297942cf67493790b18a54f0 info: Received result for tool: 07679fc0afab3f36 extract_dynamic_variable_70f8bac8 Result: {"status":"dynamic variables extracted successfully","variables":{"update_case_number":"REQ-242"}} 2026-01-05 13:10:26.798 call_184297942cf67493790b18a54f0 error: Error encountered in LLM call: Failed to get response from any LLM provider: {"error":{"message":"{\n \"error\": {\n \"code\": 400,\n \"message\": \"Unable to submit request because function call

default_api:extract_dynamic_variable_70f8bac8

in the 7. content block is missing a

thought_signature

. Learn more: https://docs.cloud.google.com/vertex-ai/generative-ai/docs/thought-signatures\",\n \"status\": \"INVALID_ARGUMENT\"\n }\n}\n","code":400,"status":"Bad Request"}} 2026-01-05 13:10:58.814 call_184297942cf67493790b18a54f0 error: Error encountered in LLM call: Failed to get response from any LLM provider: {"error":{"message":"{\n \"error\": {\n \"code\": 400,\n \"message\": --- Another example with Transfer Call: 2026-01-05 13:06:31.571 call_7f6a5309c8c188223997a8753f7 info: Received result for tool: 9553de7e23621827 transfer_call_122cde4b Result: {"status":"Call connected but no human response detected.. Please inform the customer that the transfer did not go ... them directly."} 2026-01-05 13:06:32.755 call_7f6a5309c8c188223997a8753f7 error: Error encountered in LLM call: Failed to get response from any LLM provider: {"error":{"message":"{\n \"error\": {\n \"code\": 400,\n \"message\": \"Unable to submit request because function call

default_api:transfer_call_122cde4b

in the 3. content block is missing a

thought_signature

. Learn more: https://docs.cloud.google.com/vertex-ai/generative-ai/docs/thought-signatures\",\n \"status\": \"INVALID_ARGUMENT\"\n }\n}\n","code":400,"status":"Bad Request"}}

Hey @Rakshit Kalra This is a known issue with Gemini Flash 3 where tool/function calls can fail due to missing thought signature validation, and it’s not related to your flow setup. For now, we recommend switching to another model .

any movement on this? this is important for our setup as Flash is giving us much better results.

Hello @Rakshit Kalra , there was a recent update for this model , please let us know if you had any recent issues with it.

Hi @Omar Ashraf | Community Lead - here's logs from a call just moments before you sent this message. https://cdn.discordapp.com/attachments/1457728811767562535/1459615306078097458/message.txt?ex=6965e66f&is=696494ef&hm=dabd445e482c2ad5a123e0079ea75deffd8654c0294a03968bd72c525d0d8f0b&

I've report it as bug , I will update you about it.

Hi! Checking in again to see if there are any updates. Pls advise.

Hello @Rakshit Kalra , it should be fixed now.

yup - fixed