Updated providers for paperclip and hindsight

cluster/applications/hindsight/values.yaml

@@ -88,7 +88,7 @@ api:
     # key -- only the served model name changed. Gemma 4 does NOT emit a
     # reasoning trace by default (confirmed live) -- simpler completion
     # parsing than DeepSeek-R1's always-on <think> blocks.
-    HINDSIGHT_API_LLM_BASE_URL: "http://astro-orbiter:8000/v1"
+    HINDSIGHT_API_LLM_BASE_URL: "http://illuminations.wdw-labs.cloud:8000/v1"
     HINDSIGHT_API_LLM_PROVIDER: "openai"
     HINDSIGHT_API_LLM_MODEL: "Gemma-4-26B-A4B-it-AWQ"
     # vLLM's max_model_len is now 65536 (up from DeepSeek's 32768, up from