[Bug]: Gemini 3 caching doesn't work #16961
Description
What happened?
Reproduction:
- install litellm (git) in a virtualenv
- make config.yaml:
model_list:
- model_name: gemini-3-pro-preview
litellm_params:
model: gemini/gemini-3-pro-preview
api_key: os.environ/GEMINI_API_KEY
litellm_settings:
master_key: os.environ/LITELLM_MASTER_KEY
drop_params: true
- run
litellm --config config.yaml(listening on port 7777) - run
ANTHROPIC_DEFAULT_SONNET_MODEL=gemini-3-pro-preview ANTHROPIC_DEFAULT_HAIKU_MODEL=gemini-3-pro-preview ANTHROPIC_BASE_URL=http://localhost:7777 ANTHROPIC_AUTH_TOKEN=$LITELLM_MASTER_KEY claude - send a long enough message to fit the minimum cached content size
- error
Relevant log output
litellm.exceptions.BadRequestError: litellm.BadRequestError: GeminiException BadRequestError - {
"error": {
"code": 400,
"message": "* CreateCachedContentRequest.cached_content.tools[0].function_declarations[22].parameters.properties[truncateMode].enum: only allowed for STRING type\n* CreateCachedContentRequest.cached_content.tools[0].function_declarations[34].parameters.properties[truncateMode].enum: only allowed for STRING type\n* CreateCachedContentRequest.cached_content.tools[0].function_declarations[40].parameters.properties[truncateMode].enum: only allowed for STRING type\n",
"status": "INVALID_ARGUMENT"
}
}
. Received Model Group=gemini-3-pro-preview
Available Model Group Fallbacks=None LiteLLM Retried: 1 times, LiteLLM Max Retries: 2Are you a ML Ops Team?
No
What LiteLLM version are you on ?
v1.80.0
Twitter / LinkedIn details
No response