Fix Gemini default: use gemini-2.0-flash-lite (confirmed free-tier working)

Live testing showed:
- gemini-1.5-flash* models → 404 on current API keys (deprecated/unavailable)
- gemini-2.0-flash-lite → reaches API, only fails when daily quota exhausted
- gemini-2.0-flash-001 → same (valid model, quota-limited)

Update default to gemini-2.0-flash-lite, fallback chain to 2.x models only,
and fix provider label in UI. Switch live Space to Groq (working now).

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

src/soci/api/routes.py

@@ -279,7 +279,7 @@ async def get_llm_providers():
     if os.environ.get("GROQ_API_KEY"):
         providers.append({"id": "groq",    "label": "Groq Llama 8B",       "icon": "⚡", "model": ""})
     if os.environ.get("GEMINI_API_KEY"):
-        providers.append({"id": "gemini",  "label": "Gemini 1.5 Flash",    "icon": "✦", "model": ""})
+        providers.append({"id": "gemini",  "label": "Gemini 2.0 Flash Lite", "icon": "✦", "model": ""})
     has_hf = bool(
         os.environ.get("HF_TOKEN")
         or os.environ.get("HUGGINGFACE_TOKEN")

src/soci/engine/llm.py

@@ -38,19 +38,19 @@ MODEL_GROQ_LLAMA_70B = "llama-3.3-70b-versatile"
 MODEL_GROQ_MIXTRAL = "mixtral-8x7b-32768"
 
 # Google Gemini model IDs (free tier via AI Studio)
-# gemini-1.5-flash is the reliable default on the OpenAI-compatible endpoint.
-# gemini-2.0-flash can be enabled via GEMINI_MODEL env var if your key supports it.
-MODEL_GEMINI_FLASH = "gemini-1.5-flash"
-MODEL_GEMINI_FLASH_FALLBACK = "gemini-1.5-flash"  # final fallback
-MODEL_GEMINI_FLASH_V2 = "gemini-2.0-flash"        # opt-in via GEMINI_MODEL env var
+# gemini-2.0-flash-lite is the reliable free-tier default on the OpenAI-compatible endpoint.
+# gemini-1.5-flash* models return 404 on current API keys — do not use them.
+MODEL_GEMINI_FLASH = "gemini-2.0-flash-lite"       # free tier, confirmed working
+MODEL_GEMINI_FLASH_FALLBACK = "gemini-2.0-flash-001"  # versioned fallback
 MODEL_GEMINI_PRO = "gemini-1.5-pro"
 
-# Models to try in order if a model is not available on the serverless endpoint
+# Fallback chain: tried in order when a model returns a not-available error
 _GEMINI_FALLBACK_CHAIN: dict[str, str] = {
-    "gemini-2.0-flash": MODEL_GEMINI_FLASH_FALLBACK,
-    "gemini-2.0-flash-exp": MODEL_GEMINI_FLASH_FALLBACK,
-    "gemini-2.0-flash-001": MODEL_GEMINI_FLASH_FALLBACK,
+    "gemini-2.0-flash":     "gemini-2.0-flash-lite",
+    "gemini-2.0-flash-exp": "gemini-2.0-flash-lite",
+    "gemini-2.0-flash-001": "gemini-2.0-flash-lite",
     "gemini-2.0-flash-lite": MODEL_GEMINI_FLASH_FALLBACK,
+    # 1.5-flash models return 404 on current API keys — skip the entire 1.5 family
 }
 
 # Keywords in any Gemini error body that indicate the model is unavailable on this endpoint