feat(setup,chat): detect actually-loaded local model via provider-specific API

For Ollama, /api/tags returns all installed models, not running ones.
Add fetch_loaded_models() using /api/ps for Ollama (and /v1/models for
LM Studio/llama.cpp, which already return only loaded models).

_show_local_model_status() now calls fetch_loaded_models() so the
setup wizard correctly shows only in-memory models for Ollama.

At chat session startup, local providers warn when the configured model
is not currently loaded, or when nothing is loaded at all.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

src/pyra/chat/session.py

@@ -25,6 +25,7 @@ from pyra.plugins.executor import ToolExecutor
 from pyra.plugins.registry import PluginRegistry
 from pyra.security.injection import scan_response
 from pyra.setup.providers import get_provider
+from pyra.setup.wizard import fetch_loaded_models
 from pyra.utils.paths import pyra_home
 
 _HISTORY_FILE = pyra_home() / ".chat_history"
@@ -176,6 +177,16 @@ def start_chat() -> None:
         "[dim]Type /help for commands, /quit to exit.[/dim]"
     )
 
+    if provider.group == "Local":
+        loaded = fetch_loaded_models(provider)
+        if not loaded:
+            render_info(f"No model currently loaded in {provider.display_name}.")
+        elif cfg.ai.model not in loaded:
+            render_info(
+                f"Model '{cfg.ai.model}' not loaded in {provider.display_name}. "
+                f"Loaded: {', '.join(loaded)}"
+            )
+
     _flags: dict = {"use_tools": True}
 
     while True:

src/pyra/setup/wizard.py

@@ -377,9 +377,26 @@ def _check_local_server(provider: Provider) -> None:
             # "retry" → loop
 
 
+def fetch_loaded_models(provider: Provider) -> list[str]:
+    """Return models currently loaded in RAM from a local provider's API."""
+    if not provider.base_url:
+        return []
+    try:
+        if provider.id == "ollama":
+            resp = httpx.get(f"{provider.base_url}/api/ps", timeout=3.0)
+            resp.raise_for_status()
+            return [m["name"] for m in resp.json().get("models", [])]
+        else:
+            resp = httpx.get(f"{provider.base_url}/models", timeout=3.0)
+            resp.raise_for_status()
+            return [m["id"] for m in resp.json().get("data", [])]
+    except Exception:
+        return []
+
+
 def _show_local_model_status(provider: Provider) -> None:
     """Print a one-line status showing which models are currently loaded."""
-    models = _fetch_local_models(provider)
+    models = fetch_loaded_models(provider)
     if not models:
         console.print("  [yellow]No model currently loaded[/yellow]")
     elif len(models) == 1: