diff --git a/strix/config/models.py b/strix/config/models.py
index 89ec38c..a0be563 100644
--- a/strix/config/models.py
+++ b/strix/config/models.py
@@ -49,6 +49,7 @@ def configure_sdk_model_defaults(settings: Settings) -> None:
     if llm.api_key:
         set_default_openai_key(llm.api_key, use_for_tracing=False)
         _configure_litellm_default("api_key", llm.api_key)
+        _mirror_api_key_to_provider_env(llm.model, llm.api_key)
     if llm.api_base:
         os.environ["OPENAI_BASE_URL"] = llm.api_base
         _configure_litellm_default("api_base", llm.api_base)
@@ -57,6 +58,22 @@ def configure_sdk_model_defaults(settings: Settings) -> None:
         set_default_openai_api("responses")
 
 
+def _mirror_api_key_to_provider_env(model_name: str | None, api_key: str) -> None:
+    if not model_name:
+        return
+    name = model_name.strip()
+    for prefix in ("litellm/", "any-llm/"):
+        if name.lower().startswith(prefix):
+            name = name[len(prefix) :]
+            break
+    if "/" not in name:
+        return
+    provider = name.split("/", 1)[0].strip().upper()
+    if not provider:
+        return
+    os.environ.setdefault(f"{provider}_API_KEY", api_key)
+
+
 def _configure_litellm_compatibility() -> None:
     """Enable LiteLLM's permissive param-handling mode."""
     import litellm