实际执行任务模型与请求模型不符合

[koast18]

似乎网关并没有尊重用户请求的模型名，服务端发回的sse里统一都是auto模型

例如

{"headers":{"Content-Type":["application/json"]},"statusCodeValue":200,"statusCode":"OK","body":"{\"choices\":[{\"delta\":{\"reasoning_content\":\"Thinki
 ng\",\"role\":\"assistant\"},\"index\":0}],\"created\":xxxxxxxxxxxxx,\"id\":\"chatcmpl-uuid\",\"model\":\"auto\",\"object\":
 \"chat.completion.chunk\"}"}

{"headers":{"Content-Type":["application/json"]},"statusCodeValue":200,"statusCode":"OK","body":"{\"choices\":[],\"created\":xxxxxxxxxxxxxx,\"id\":\"chatcmpl
 -uuid\",\"model\":\"auto\",\"object\":\"chat.completion.chunk\",\"usage\":{\"billable\":true,\"completion_tokens\":xxx,\"
 completion_tokens_details\":{\"reasoning_tokens\":xxx},\"credits\":0.59xx,\"original_credits\":0.59xx,\"prompt_tokens\":xx,\"prompt_tokens_details\":{},\
 "total_tokens\":xxx}}"}

而且返回模型的思维链很像是Qwen，感觉只要是请求api3就会是这样子。