diff --git a/index/main.py b/index/main.py
index 5ae3343..d6083ea 100644
--- a/index/main.py
+++ b/index/main.py
@@ -1,7 +1,6 @@
 import asyncio
 import logging
 import os
-from contextlib import asynccontextmanager
 from functools import lru_cache
 from typing import Any
 
@@ -74,7 +73,7 @@ class SparseVector(BaseModel):
     values: list[float]
 
 
-CHUNK_SIZE = 384
+CHUNK_SIZE = 256
 OVERLAP_SIZE = 128
 SPARSE_MODEL_NAME = "Qdrant/bm25"
 FASTEMBED_CACHE_PATH = "/models/fastembed"
@@ -102,9 +101,6 @@ def render_message(message: Message) -> str:
                 else:
                     parts_list.append(part_text)
 
-    if message.mentions:
-        parts_list.append(" ".join(message.mentions))
-
     if message.file_snippets:
         parts_list.append(f"[Файл]: {message.file_snippets}")
 
@@ -190,13 +186,7 @@ def build_chunks(
     return result
 
 
-@asynccontextmanager
-async def lifespan(app: FastAPI):
-    await asyncio.to_thread(get_sparse_model)
-    yield
-
-
-app = FastAPI(title="Index Service", version="0.1.0", lifespan=lifespan)
+app = FastAPI(title="Index Service", version="0.1.0")
 
 
 @app.get("/health")
diff --git a/search/main.py b/search/main.py
index 7f9facb..763658d 100644
--- a/search/main.py
+++ b/search/main.py
@@ -148,10 +148,11 @@ def get_sparse_model() -> SparseTextEmbedding:
 
 @asynccontextmanager
 async def lifespan(app: FastAPI):
-    await asyncio.to_thread(get_sparse_model)
-    limits = httpx.Limits(max_connections=100, max_keepalive_connections=20)
-    app.state.http = httpx.AsyncClient(timeout=30.0, limits=limits)
-    app.state.qdrant = AsyncQdrantClient(url=QDRANT_URL, api_key=API_KEY)
+    app.state.http = httpx.AsyncClient()
+    app.state.qdrant = AsyncQdrantClient(
+        url=QDRANT_URL,
+        api_key=API_KEY,
+    )
     try:
         yield
     finally:
@@ -161,10 +162,10 @@ async def lifespan(app: FastAPI):
 
 app = FastAPI(title="Search Service", version="0.1.0", lifespan=lifespan)
 
-DENSE_PREFETCH_K = 50
-SPARSE_PREFETCH_K = 150
-RETRIEVE_K = 100
-RERANK_LIMIT = 15
+DENSE_PREFETCH_K = 80
+SPARSE_PREFETCH_K = 200
+RETRIEVE_K = 150
+RERANK_LIMIT = 10
 
 
 async def embed_dense(client: httpx.AsyncClient, text: str) -> list[float]:
@@ -353,7 +354,7 @@ async def search(payload: SearchAPIRequest) -> SearchAPIResponse:
     sparse_query = build_sparse_query(question)
 
     dense_task = embed_dense(client, dense_query)
-    sparse_task = asyncio.to_thread(embed_sparse_sync, sparse_query)
+    sparse_task = asyncio.to_thread(lambda: embed_sparse_sync(sparse_query))
     dense_vector, sparse_vector = await asyncio.gather(dense_task, sparse_task)
 
     dense_vectors = [dense_vector]