litellm: shadow gpt-4o-mini / gpt-4o / gpt-4-turbo aliases onto Claude backends

paperless-ai's setup wizard validates the OpenAI provider by hardcoding
model=gpt-4o-mini in the probe, regardless of the OPENAI_MODEL env. Without
the alias LiteLLM 400s ("Invalid model name") and the wizard rejects the
key. Shadow common OpenAI names onto our Claude backends so any client that
probes gpt-* gets a healthy response (and routes to the Max sub).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

vars/main.yml

@@ -43,15 +43,26 @@ litellm_port: 4000
 litellm_host: "0.0.0.0"
 litellm_package_spec: "litellm[proxy]==1.55.10"
 # Models map onto Meridian's pinned Anthropic-shape backend on 127.0.0.1:3456.
-# Add Sonnet/Opus aliases too so clients can pick cost vs quality without
+# Native Claude aliases AND OpenAI-named aliases — some clients (paperless-ai's
-# touching this config.
+# setup wizard) hardcode `model=gpt-4o-mini` for validation regardless of
+# what you set as the default model, so we shadow the common OpenAI names too.
 litellm_models:
+  # Native Claude aliases (preferred for new clients)
   - name: claude-haiku-4-5
     backend: anthropic/claude-haiku-4-5
   - name: claude-sonnet-4-6
     backend: anthropic/claude-sonnet-4-6
   - name: claude-opus-4-7
     backend: anthropic/claude-opus-4-7
+  # OpenAI-name shadows — for clients that probe gpt-* names regardless of
+  # config (paperless-ai wizard, Open WebUI defaults, etc.). All actually
+  # backed by Claude on the Max sub.
+  - name: gpt-4o-mini
+    backend: anthropic/claude-haiku-4-5
+  - name: gpt-4o
+    backend: anthropic/claude-sonnet-4-6
+  - name: gpt-4-turbo
+    backend: anthropic/claude-sonnet-4-6
 # Master key is required by LiteLLM. Pulled at deploy time from Infisical
 # /meridian/vault_litellm_master_key and passed via -e on the playbook
 # (see deploy.sh).