Update app.py

app.py

@@ -65,23 +65,41 @@ STREAM_RESPONSE_HEADERS = {
 # MODELS
 # =========================================================
 
+# Pool of free models — rotated automatically on 429
+FREE_MODELS_POOL = [
+    "deepseek-v4-flash-free",
+    "qwen3.6-plus-free",
+    "minimax-m2.5-free",
+    "nemotron-3-super-free",
+    "big-pickle",
+]
+
 MODEL_MAP = {
-    "claude-sonnet-4-6": "deepseek-v4-flash-free",
-    "claude-sonnet-4-5": "deepseek-v4-flash-free",
-    "claude-haiku-4-6": "deepseek-v4-flash-free",
-    "claude-haiku-4-5": "deepseek-v4-flash-free",
+    "claude-sonnet-4-6":         "deepseek-v4-flash-free",
+    "claude-sonnet-4-5":         "deepseek-v4-flash-free",
+    "claude-haiku-4-6":          "deepseek-v4-flash-free",
+    "claude-haiku-4-5":          "deepseek-v4-flash-free",
     "claude-haiku-4-5-20251001": "deepseek-v4-flash-free",
-    "claude-opus-4-6": "nemotron-3-super-free",
-
-    "deepseek-v4-flash-free": "deepseek-v4-flash-free",
-    "nemotron-3-super-free": "nemotron-3-super-free",
-    "qwen3.6-plus-free": "qwen3.6-plus-free",
-    "minimax-m2.5-free": "minimax-m2.5-free",
-    "big-pickle": "big-pickle",
+    "claude-opus-4-6":           "nemotron-3-super-free",
+    "deepseek-v4-flash-free":    "deepseek-v4-flash-free",
+    "nemotron-3-super-free":     "nemotron-3-super-free",
+    "qwen3.6-plus-free":         "qwen3.6-plus-free",
+    "minimax-m2.5-free":         "minimax-m2.5-free",
+    "big-pickle":                "big-pickle",
 }
 
 DEFAULT_MODEL = "deepseek-v4-flash-free"
 
+
+def _next_model(current_model):
+    """Return the next free model in the pool after current_model."""
+    try:
+        idx = FREE_MODELS_POOL.index(current_model)
+    except ValueError:
+        idx = -1
+    next_idx = (idx + 1) % len(FREE_MODELS_POOL)
+    return FREE_MODELS_POOL[next_idx]
+
 # =========================================================
 # BROWSER TOOLS FILTER
 # =========================================================
@@ -1316,9 +1334,17 @@ async def messages(
 
                     last_error = raw_error
 
-                    # 429 = rate limit: stop immediately
+                    # 429 = rate limit: rotate to next free model
                     if response.status_code == 429:
-                        break
+                        next_m = _next_model(mapped_model)
+                        logger.warning(
+                            "[%s] 429 on %s -> rotating to %s",
+                            request_id, mapped_model, next_m,
+                        )
+                        mapped_model = next_m
+                        payload["model"] = mapped_model
+                        await asyncio.sleep(1)
+                        continue
 
                     await asyncio.sleep(RETRY_DELAY)
 
@@ -1353,9 +1379,17 @@ async def messages(
 
                 last_error = response.text
 
-                # 429 = rate limit: stop immediately
+                # 429 = rate limit: rotate to next free model
                 if response.status_code == 429:
-                    break
+                    next_m = _next_model(mapped_model)
+                    logger.warning(
+                        "[%s] 429 on %s -> rotating to %s",
+                        request_id, mapped_model, next_m,
+                    )
+                    mapped_model = next_m
+                    payload["model"] = mapped_model
+                    await asyncio.sleep(1)
+                    continue
 
                 await asyncio.sleep(RETRY_DELAY)
 
@@ -1403,9 +1437,9 @@ async def messages(
             await asyncio.sleep(RETRY_DELAY)
 
     return JSONResponse(
-        status_code=429 if "429" in str(last_error) else 502,
+        status_code=502,
         content={
-            "error": "Rate limited" if "429" in str(last_error) else "Upstream failed",
-            "details": last_error
+            "error": "All free models exhausted or upstream failed",
+            "details": last_error,
         },
     )