Upload README.md

README.md

@@ -25,10 +25,29 @@ secrets:
 
 # Space Multiverse — ML Intern V2 (FREE & SAVE MODEL)
 
-<!-- rebuild: 2026-06-05-fixed-owl-alpha -->
+<!-- rebuild: 2026-06-06-owl-alpha-removed-stealth-broken -->
 
-* Fixed: Owl Alpha now uses `openai/openrouter/owl-alpha` (NO `:free` suffix — it's free by default).
-* Fixed: All other OpenRouter free models use `:free` suffix.
-* Fixed: OpenRouter models are routed to `https://openrouter.ai/api/v1` with stripped `openai/` prefix.
-* Fixed: Frontend dropdown shows 13 custom models instead of upstream defaults.
-* Required: Set `OPENROUTER_API_KEY` secret in Space settings.
+**13 models — mix of HF Inference + OpenRouter.**
+
+## Models
+
+| # | Model | Route | Provider | Free? |
+|---|-------|-------|----------|-------|
+| 1 | **DeepSeek V4 Pro** | HF Inference | `deepseek-ai` | ✅ (default) |
+| 2 | **DeepSeek V4 Flash** | HF Inference | `deepseek-ai` | ✅ |
+| 3 | **GPT-5.5** | HF Inference (fal-ai) | `openai` | ✅ |
+| 4 | **Qwen3 Coder Next** | HF Inference | `Qwen` | ✅ |
+| 5 | **Gemma 3 1B** | HF Inference | `google` | ✅ |
+| 6 | **Gemini 2.0 Flash** | OpenRouter | `google` | ✅ |
+| 7 | **DeepSeek V4 Flash (OR)** | OpenRouter | `deepseek` | ✅ |
+| 8 | **Gemma 4 9B** | OpenRouter | `google` | ✅ |
+| 9 | **Gemma 4 31B** | OpenRouter | `google` | ✅ |
+| 10 | **Nemotron 3 Super 120B** | OpenRouter | `nvidia` | ✅ |
+| 11 | **Nemotron 3 Ultra 550B** | OpenRouter | `nvidia` | ✅ |
+| 12 | **Laguna M.1** | OpenRouter | `poolside` | ✅ |
+| 13 | **Laguna XS.2** | OpenRouter | `poolside` | ✅ |
+
+> Owl Alpha REMOVED — its free provider "Stealth" on OpenRouter is broken (returns 400 error).
+
+## Required Secrets
+- `OPENROUTER_API_KEY` — required for OpenRouter-routed models