From 072214c7bb1f44db76d7f3c21da91e5d4791ed46 Mon Sep 17 00:00:00 2001 From: huanghaiyang Date: Wed, 8 Apr 2026 14:43:55 +0800 Subject: [PATCH] fix(proxy): - Fixed issue where Volcengine embedding models were incorrectly identified as "text-embedding-3-small" - Implemented route type-based model selection in proxy server - Use embedding_model setting for embedding requests instead of completion_model - Ensure user-provided model names take precedence over default settings - Maintain compatibility with completion requests using completion_model setting --- litellm/proxy/common_request_processing.py | 21 +++++++++++++++------ 1 file changed, 15 insertions(+), 6 deletions(-) diff --git a/litellm/proxy/common_request_processing.py b/litellm/proxy/common_request_processing.py index 09200c96841..17876a088b0 100644 --- a/litellm/proxy/common_request_processing.py +++ b/litellm/proxy/common_request_processing.py @@ -700,12 +700,21 @@ class ProxyBaseLLMRequestProcessing: "queue_time_seconds" ] = queue_time_seconds - self.data["model"] = ( - general_settings.get("completion_model", None) # server default - or user_model # model name passed via cli args - or model # for azure deployments - or self.data.get("model", None) # default passed in http request - ) + # Use appropriate model setting based on route type + if route_type == "aembedding": + self.data["model"] = ( + general_settings.get("embedding_model", None) # server default for embedding + or user_model # model name passed via cli args + or model # for azure deployments + or self.data.get("model", None) # default passed in http request + ) + else: + self.data["model"] = ( + general_settings.get("completion_model", None) # server default for completion + or user_model # model name passed via cli args + or model # for azure deployments + or self.data.get("model", None) # default passed in http request + ) # override with user settings, these are params passed via cli if user_temperature: