Move runner port 18008 -> 18010

Follows the change already made on 4n4rch02. Updates the compose --port and
healthcheck plus every default in the bench tooling and the README, so the
scripts keep working without an explicit --port.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

README.md

@@ -2,7 +2,7 @@
 
 llama.cpp CUDA runner for **Qwen3.6-35B-A3B** (UD-Q4_K_M) at **128k context** on 4n4rch02.
 
-Port **18008**, OpenAI-compatible API at `http://192.168.3.189:18008/v1`.
+Port **18010**, OpenAI-compatible API at `http://192.168.3.189:18010/v1`.
 
 ## Host
 
@@ -262,7 +262,7 @@ Loads in ~15 s (`--no-mmap`, weights read from NVMe).
 ## Benchmarks
 
 ```bash
-python3 bench/bench.py --port 18008 --depths 512,8192,32768 --n-predict 128
+python3 bench/bench.py --port 18010 --depths 512,8192,32768 --n-predict 128
 ./bench/sweep.sh 10:24,16:8 8:24,16:8      # <ncmoe>:<tensor-split>:<threads>
 python3 bench/spec_bench.py --no-think     # copy/code/prose generation workloads
 ```

bench/bench.py

@@ -40,7 +40,7 @@ def bench(port, prompt_tokens, n_predict, label):
 
 if __name__ == "__main__":
     ap = argparse.ArgumentParser()
-    ap.add_argument("--port", type=int, default=18008)
+    ap.add_argument("--port", type=int, default=18010)
     ap.add_argument("--label", default="run")
     ap.add_argument("--depths", default="512,8192,32768,131072")
     ap.add_argument("--n-predict", type=int, default=128)

bench/reasoning_budget_probe.py

@@ -53,7 +53,7 @@ def probe(port, label, extra):
 
 if __name__ == "__main__":
     ap = argparse.ArgumentParser()
-    ap.add_argument("--port", type=int, default=18008)
+    ap.add_argument("--port", type=int, default=18010)
     ap.add_argument("--case", choices=["default", "override", "off"], default="default")
     ap.add_argument("--budget", type=int, default=16384)
     a = ap.parse_args()

bench/spec_bench.py

@@ -61,7 +61,7 @@ def run(port, name, prompt, n_predict, no_think):
 
 if __name__ == "__main__":
     ap = argparse.ArgumentParser()
-    ap.add_argument("--port", type=int, default=18008)
+    ap.add_argument("--port", type=int, default=18010)
     ap.add_argument("--label", default="run")
     ap.add_argument("--n-predict", type=int, default=400)
     ap.add_argument("--no-think", action="store_true",

bench/sweep.sh

@@ -4,12 +4,12 @@
 # Usage:   ./sweep.sh <ncmoe>:<tensor-split>[:<threads>] ...
 # Example: ./sweep.sh 10:24,16:8 8:24,16:8
 #
-# Env: CTX (default 131072), DEPTHS (default 512,8192), PORT (default 18008)
+# Env: CTX (default 131072), DEPTHS (default 512,8192), PORT (default 18010)
 
 HUB=${HUB:-/mnt/2TSAM990nvme/docker-volume-outsource/llm-models/hf/hub/models--unsloth--Qwen3.6-35B-A3B-GGUF}
 SNAP=${SNAP:-snapshots/a483e9e6cbd595906af30beda3187c2663a1118c/Qwen3.6-35B-A3B-UD-Q4_K_M.gguf}
 CTX=${CTX:-131072}
-PORT=${PORT:-18008}
+PORT=${PORT:-18010}
 NAME=qwen36-sweep
 BENCH=${BENCH:-$(dirname "$0")/bench.py}
 

docker-compose.yml

@@ -72,9 +72,9 @@ services:
       - --host
       - 0.0.0.0
       - --port
-      - "18008"
+      - "18010"
     healthcheck:
-      test: ["CMD", "curl", "-sf", "http://localhost:18008/health"]
+      test: ["CMD", "curl", "-sf", "http://localhost:18010/health"]
       interval: 30s
       timeout: 10s
       retries: 3