bug: ai plugins log summaries and payloads even when logging options are set to false

[mikyll]

Current Behavior

When using ai-proxy (and similarly ai-proxy-multi), setting:

• logging.summaries = false
• logging.payloads = false

does not fully disable AI request logging at info level in APISIX runtime logs (error.log).

I still see request-related logs emitted from the AI driver at INFO level, including request metadata and potentially payload-bearing structures (e.g. API key and Authorization: Bearer header).

This, even though is not a issue with higher log levels than info, differs from what stated in documentation.

Expected Behavior

According to ai-proxy documentation, logging.summaries and logging.payloads are set to false by default, and information about the requests should not be logged.

• logging.summaries:

  If true, logs request LLM model, duration, request, and response tokens.

• logging.payloads:

  If true, logs request and response payload.

Error Logs

With logging.summaries = true and logging.payloads = true:

2026/03/25 17:14:10 [info] 89#89: *2112 [lua] trusted-addresses.lua:46: is_trusted(): trusted_addresses_matcher is not initialized, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 [lua] ai.lua:243: handler(): use ai plane to match route, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 [lua] ai.lua:247: handler(): renew route cache: count=3001, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 [lua] radixtree.lua:493: common_route_data(): path: /chat/completions operator: =, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 [lua] ai.lua:77: match(): route match mode: ai_match, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 [lua] ai.lua:80: match(): route cache key: /chat/completions, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 [lua] radixtree_host_uri.lua:161: orig_router_http_matching(): route match mode: radixtree_host_uri, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 [lua] init.lua:733: http_access_phase(): matched route: {"clean_handlers":{},"orig_modifiedIndex":1774458846,"has_domain":false,"modifiedIndex":1774458846,"key":"/routes/ai_endpoint","value":{"plugins":{"ai-proxy":{"_meta":{},"keepalive_pool":30,"auth":{"header":{"Authorization":"Bearer myapikey"}},"logging":{"payloads":true,"summaries":true},"ssl_verify":true,"options":{"model":"gemini-2.5-flash"},"provider":"openai-compatible","override":{"endpoint":"https://generativelanguage.googleapis.com/v1beta/openai/chat/completions"},"timeout":30000,"keepalive_timeout":60000,"keepalive":true}},"uri":"/chat/completions","priority":0,"id":"ai_endpoint","status":1}}, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 [lua] openai-base.lua:258: request extra_opts to LLM server: {"model_options":{"model":"gemini-2.5-flash"},"endpoint":"https://generativelanguage.googleapis.com/v1beta/openai/chat/completions","auth":{"header":{"Authorization":"Bearer myapikey"}},"conf":{}}, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 [lua] openai-base.lua:336: sending request to LLM server: {"method":"POST","scheme":"https","ssl_verify":true,"headers":{"Authorization":"Bearer myapikey","Content-Type":"application/json"},"query":{},"port":443,"ssl_server_name":"generativelanguage.googleapis.com","host":"generativelanguage.googleapis.com","path":"/v1beta/openai/chat/completions","body":{"messages":[{"content":"Explain to me how AI works","role":"user"}],"model":"gemini-2.5-flash"}}, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 [lua] client.lua:123: dns_parse(): dns resolve generativelanguage.googleapis.com, result: {"type":1,"name":"generativelanguage.googleapis.com","address":"172.217.23.74","ttl":96,"class":1,"section":1}, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 [lua] resolver.lua:84: parse_domain(): parse addr: {"type":1,"name":"generativelanguage.googleapis.com","address":"172.217.23.74","ttl":96,"class":1,"section":1}, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 [lua] resolver.lua:85: parse_domain(): resolver: ["10.89.29.1"], client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 [lua] resolver.lua:86: parse_domain(): host: generativelanguage.googleapis.com, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 [lua] resolver.lua:88: parse_domain(): dns resolver domain: generativelanguage.googleapis.com to 172.217.23.74, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 [lua] openai-base.lua:186: read_response(): got token usage from ai service: null, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 client 10.89.29.187 closed keepalive connection
[apisix]  | 10.89.29.187 - - [25/Mar/2026:17:14:10 +0000] localhost:9080 "POST /chat/completions HTTP/1.1" 400 595 0.111 "-" "curl/8.5.0" - - - "http://localhost:9080"

With logging.summaries = false and logging.payloads = false:

2026/03/25 17:14:40 [info] 63#63: *11948 [lua] trusted-addresses.lua:46: is_trusted(): trusted_addresses_matcher is not initialized, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "4a65c35421c5a84506a1375d4f8b24a8"
2026/03/25 17:14:40 [info] 63#63: *11948 [lua] ai.lua:243: handler(): use ai plane to match route, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "4a65c35421c5a84506a1375d4f8b24a8"
2026/03/25 17:14:40 [info] 63#63: *11948 [lua] ai.lua:247: handler(): renew route cache: count=3001, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "4a65c35421c5a84506a1375d4f8b24a8"
2026/03/25 17:14:40 [info] 63#63: *11948 [lua] radixtree.lua:493: common_route_data(): path: /chat/completions operator: =, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "4a65c35421c5a84506a1375d4f8b24a8"
2026/03/25 17:14:40 [info] 63#63: *11948 [lua] ai.lua:77: match(): route match mode: ai_match, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "4a65c35421c5a84506a1375d4f8b24a8"
2026/03/25 17:14:40 [info] 63#63: *11948 [lua] ai.lua:80: match(): route cache key: /chat/completions, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "4a65c35421c5a84506a1375d4f8b24a8"
2026/03/25 17:14:40 [info] 63#63: *11948 [lua] radixtree_host_uri.lua:161: orig_router_http_matching(): route match mode: radixtree_host_uri, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "4a65c35421c5a84506a1375d4f8b24a8"
2026/03/25 17:14:40 [info] 63#63: *11948 [lua] init.lua:733: http_access_phase(): matched route: {"clean_handlers":{},"orig_modifiedIndex":1774458877,"has_domain":false,"modifiedIndex":1774458877,"key":"/routes/ai_endpoint","value":{"plugins":{"ai-proxy":{"_meta":{},"keepalive_pool":30,"auth":{"header":{"Authorization":"Bearer myapikey"}},"logging":{"payloads":false,"summaries":false},"ssl_verify":true,"options":{"model":"gemini-2.5-flash"},"provider":"openai-compatible","override":{"endpoint":"https://generativelanguage.googleapis.com/v1beta/openai/chat/completions"},"timeout":30000,"keepalive_timeout":60000,"keepalive":true}},"uri":"/chat/completions","priority":0,"id":"ai_endpoint","status":1}}, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "4a65c35421c5a84506a1375d4f8b24a8"
2026/03/25 17:14:40 [info] 63#63: *11948 [lua] openai-base.lua:258: request extra_opts to LLM server: {"model_options":{"model":"gemini-2.5-flash"},"endpoint":"https://generativelanguage.googleapis.com/v1beta/openai/chat/completions","auth":{"header":{"Authorization":"Bearer myapikey"}},"conf":{}}, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "4a65c35421c5a84506a1375d4f8b24a8"
2026/03/25 17:14:40 [info] 63#63: *11948 [lua] openai-base.lua:336: sending request to LLM server: {"method":"POST","scheme":"https","ssl_verify":true,"headers":{"Authorization":"Bearer myapikey","Content-Type":"application/json"},"query":{},"port":443,"ssl_server_name":"generativelanguage.googleapis.com","host":"generativelanguage.googleapis.com","path":"/v1beta/openai/chat/completions","body":{"messages":[{"content":"Explain to me how AI works","role":"user"}],"model":"gemini-2.5-flash"}}, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "4a65c35421c5a84506a1375d4f8b24a8"
2026/03/25 17:14:40 [info] 63#63: *11948 [lua] client.lua:123: dns_parse(): dns resolve generativelanguage.googleapis.com, result: {"type":1,"name":"generativelanguage.googleapis.com","address":"172.217.23.170","ttl":66,"class":1,"section":1}, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "4a65c35421c5a84506a1375d4f8b24a8"
2026/03/25 17:14:40 [info] 63#63: *11948 [lua] resolver.lua:84: parse_domain(): parse addr: {"type":1,"name":"generativelanguage.googleapis.com","address":"172.217.23.170","ttl":66,"class":1,"section":1}, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "4a65c35421c5a84506a1375d4f8b24a8"
2026/03/25 17:14:40 [info] 63#63: *11948 [lua] resolver.lua:85: parse_domain(): resolver: ["10.89.29.1"], client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "4a65c35421c5a84506a1375d4f8b24a8"
2026/03/25 17:14:40 [info] 63#63: *11948 [lua] resolver.lua:86: parse_domain(): host: generativelanguage.googleapis.com, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "4a65c35421c5a84506a1375d4f8b24a8"
2026/03/25 17:14:40 [info] 63#63: *11948 [lua] resolver.lua:88: parse_domain(): dns resolver domain: generativelanguage.googleapis.com to 172.217.23.170, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "4a65c35421c5a84506a1375d4f8b24a8"
2026/03/25 17:14:40 [info] 63#63: *11948 [lua] openai-base.lua:186: read_response(): got token usage from ai service: null, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "4a65c35421c5a84506a1375d4f8b24a8"
2026/03/25 17:14:40 [info] 63#63: *11948 client 10.89.29.187 closed keepalive connection
[apisix]  | 10.89.29.187 - - [25/Mar/2026:17:14:40 +0000] localhost:9080 "POST /chat/completions HTTP/1.1" 400 595 0.119 "-" "curl/8.5.0" - - - "http://localhost:9080"

Notice how we see the following entries in both cases:

2026/03/25 17:14:10 [info] 89#89: *2112 [lua] openai-base.lua:258: request extra_opts to LLM server: {"model_options":{"model":"gemini-2.5-flash"},"endpoint":"https://generativelanguage.googleapis.com/v1beta/openai/chat/completions","auth":{"header":{"Authorization":"Bearer myapikey"}},"conf":{}}, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"
2026/03/25 17:14:10 [info] 89#89: *2112 [lua] openai-base.lua:336: sending request to LLM server: {"method":"POST","scheme":"https","ssl_verify":true,"headers":{"Authorization":"Bearer myapikey","Content-Type":"application/json"},"query":{},"port":443,"ssl_server_name":"generativelanguage.googleapis.com","host":"generativelanguage.googleapis.com","path":"/v1beta/openai/chat/completions","body":{"messages":[{"content":"Explain to me how AI works","role":"user"}],"model":"gemini-2.5-flash"}}, client: 10.89.29.187, server: _, request: "POST /chat/completions HTTP/1.1", host: "localhost:9080", request_id: "a96099056c57a2dfe01abd8b776708c8"

Steps to Reproduce

MRE (APISIX standalone):

• File compose.yaml:

  services:
    apisix:
      container_name: apisix
      image: apache/apisix:3.15.0-ubuntu
      volumes:
        - ./apisix/conf/config.yaml:/usr/local/apisix/conf/config.yaml
        - ./apisix/conf/apisix.yaml:/usr/local/apisix/conf/apisix.yaml
      ports:
        - "9080:9080/tcp"
        - "9443:9443/tcp"
      networks:
        - apisix
  
    httpbin:
      container_name: httpbin
      image: kennethreitz/httpbin:latest
      networks:
        - apisix
  
  networks:
    apisix:
      driver: bridge

• File conf/config.yaml:

  deployment:
    role: data_plane
    role_data_plane:
      config_provider: yaml
  
  nginx_config:
    error_log_level:  info

• File conf/apisix.yaml:

  upstreams:
    - id: httpbin
      nodes:
        "httpbin:80": 1
      type: roundrobin
  
  routes:
    - id: ai_endpoint
      uri: /chat/completions
      plugins:
        ai-proxy:
          provider: openai-compatible
          auth:
            header:
              Authorization: Bearer myapikey
          options:
            model: gemini-2.5-flash
          override:
            endpoint: https://generativelanguage.googleapis.com/v1beta/openai/chat/completions
  
  #END

Run the containers (I use podman):

podman-compose up

Send a request (ignore the 400 which isn't relevant to this example):

curl "localhost:9080/chat/completions" -H 'Content-Type: application/json' -d '{"messages":[{"role":"user","content":"Explain to me how AI works"}]}'

Have a look at the logs.

Environment

• APISIX version (run apisix version): 3.15.0
• Operating system (run uname -a): Ubuntu 24.04

[Baoyuantop]

Hi @mikyll, thank you for reporting this issue. There are two issues here:

1. The key was printed in the INFO log, which is not allowed; we will fix this.
2. You have misunderstood the functions of these two configuration options. logging.summaries and logging.payloads are designed to control whether LLM request summaries and payloads are appended to structured access log entries (which are consumed by logging plugins such as http-logger and kafka-logger). They were never intended to control core.log.info debug logs in the error.log. The term “logs” in the documentation is indeed ambiguous—it does not specify whether it refers to “structured access logs” or “error.log debug logs.” We will update this section of the documentation.