Update app.py

app.py

@@ -33,31 +33,14 @@ SGLANG_BASE = os.getenv("DARWIN_API", "http://localhost:7947")
 SGLANG_URL  = f"{SGLANG_BASE}/v1/chat/completions"
 
 # Multi-model config
-MODELS = {
-    "Darwin-35B-A3B-Opus": {
-        "id": "FINAL-Bench/Darwin-35B-A3B-Opus",
-        "api": os.getenv("DARWIN_API", "http://localhost:7947"),
-        "arch": "MoE", "active": "3B / 35B total",
-        "ctx": "262K", "thinking": True, "vision": True,
-        "max_tokens": 16384, "temp_max": 1.5,
-        "desc": "Original BF16 · SGLang",
-        "badge": "BF16",
-    },
-    "Darwin-35B-A3B-Opus-Q8-GGUF": {
-        "id": "FINAL-Bench/Darwin-35B-A3B-Opus-Q8-GGUF",
-        "api": os.getenv("DARWIN_GGUF_API", "http://localhost:8080"),
-        "arch": "MoE", "active": "3B / 35B total",
-        "ctx": "262K", "thinking": True, "vision": False,
-        "max_tokens": 16384, "temp_max": 1.5,
-        "desc": "Q8_0 GGUF · llama.cpp",
-        "badge": "Q8 GGUF",
-    },
+MODEL_NAME = "Darwin-35B-A3B-Opus"
+MODEL_ID   = "FINAL-Bench/Darwin-35B-A3B-Opus"
+MODEL_CAP  = {
+    "arch": "MoE", "active": "3B / 35B total",
+    "ctx": "262K", "thinking": True, "vision": True,
+    "max_tokens": 16384, "temp_max": 1.5,
 }
 
-DEFAULT_MODEL = "Darwin-35B-A3B-Opus"
-MODEL_NAME = DEFAULT_MODEL
-MODEL_CAP  = MODELS[DEFAULT_MODEL]
-
 PRESETS = {
     "general":   "You are Darwin-35B-A3B-Opus, a highly capable reasoning model created by VIDRAFT via evolutionary merge. Think step by step for complex questions.",
     "code":      "You are an expert software engineer. Write clean, efficient, well-commented code. Explain your approach before writing. Use modern best practices.",
@@ -174,14 +157,12 @@ def generate_reply(
     max_new_tokens: int,
     temperature:    float,
     top_p:          float,
-    model_name:     str = "Darwin-35B-A3B-Opus",
 ) -> Generator[str, None, None]:
 
-    mcfg = MODELS.get(model_name, MODELS[DEFAULT_MODEL])
-    api_url = f"{mcfg['api']}/v1/chat/completions"
+    api_url = f"{SGLANG_BASE}/v1/chat/completions"
     use_think = "Thinking" in thinking_mode
-    max_new_tokens = min(int(max_new_tokens), mcfg["max_tokens"])
-    temperature    = min(float(temperature),  mcfg["temp_max"])
+    max_new_tokens = min(int(max_new_tokens), MODEL_CAP["max_tokens"])
+    temperature    = min(float(temperature),  MODEL_CAP["temp_max"])
 
     messages: list[dict] = []
     if system_prompt.strip():
@@ -218,7 +199,7 @@ def generate_reply(
     user_text = build_user_message(message, use_think)
 
     # Vision: image input handling
-    if image_input and mcfg["vision"]:
+    if image_input and MODEL_CAP["vision"]:
         import io
         from PIL import Image as PILImage
 
@@ -247,7 +228,7 @@ def generate_reply(
         api_urls.append(f"{H100_API.rstrip('/')}/v1/chat/completions")
 
     request_body = {
-        "model": mcfg["id"],
+        "model": MODEL_ID,
         "messages": messages,
         "max_tokens": max_new_tokens,
         "temperature": temperature,
@@ -327,7 +308,6 @@ with gr.Blocks(title="Darwin-35B-A3B-Opus") as gradio_demo:
     max_new_tokens = gr.Slider(minimum=64, maximum=16384, value=4096, visible=False)
     temperature    = gr.Slider(minimum=0.0, maximum=1.5, value=0.6,  visible=False)
     top_p          = gr.Slider(minimum=0.1, maximum=1.0, value=0.9,  visible=False)
-    model_selector = gr.Textbox(value=DEFAULT_MODEL, visible=False)
 
     gr.ChatInterface(
         fn=generate_reply,
@@ -335,7 +315,6 @@ with gr.Blocks(title="Darwin-35B-A3B-Opus") as gradio_demo:
         additional_inputs=[
             thinking_toggle, image_input,
             system_prompt, max_new_tokens, temperature, top_p,
-            model_selector,
         ],
     )
 
@@ -433,14 +412,6 @@ async def health():
     except:
         return {"status":"ok","sglang":"disconnected"}
 
-@fapp.get("/api/models")
-async def api_models():
-    return JSONResponse({
-        "models": {k: {"desc": v["desc"], "badge": v["badge"], "vision": v["vision"], "ctx": v["ctx"]}
-                   for k, v in MODELS.items()},
-        "default": DEFAULT_MODEL,
-    })
-
 # ── Web Search API (Brave) ──────────────────────────────────────────────
 BRAVE_API_KEY = os.getenv("BRAVE_API_KEY", "")