diff --git a/devplacepy/main.py b/devplacepy/main.py
index dcaa7787..1d861c6c 100644
--- a/devplacepy/main.py
+++ b/devplacepy/main.py
@@ -12,7 +12,6 @@ from fastapi import FastAPI, Request
 from fastapi.responses import HTMLResponse, RedirectResponse
 from fastapi.staticfiles import StaticFiles
 from fastapi.exceptions import RequestValidationError
-from starlette.middleware.gzip import GZipMiddleware
 from devplacepy.config import (
     STATIC_DIR,
     STATIC_VERSION,
@@ -610,10 +609,6 @@ async def response_timing(request: Request, call_next):
     response.headers["X-Response-Time"] = f"{(time.perf_counter() - start) * 1000:.1f}ms"
     return response
 
-
-app.add_middleware(GZipMiddleware, minimum_size=512, compresslevel=5)
-
-
 _home_cache = TTLCache(ttl=int(os.environ.get("DEVPLACE_HOME_CACHE_TTL", "60")), max_size=4)
 
 
diff --git a/devplacepy/templates/docs/architecture-backend.html b/devplacepy/templates/docs/architecture-backend.html
index 2fa6b10a..4d6aadcf 100644
--- a/devplacepy/templates/docs/architecture-backend.html
+++ b/devplacepy/templates/docs/architecture-backend.html
@@ -11,7 +11,7 @@ How a request flows through middleware to a router, how the database layer is bu
 
 ## Middleware
 
-Seven HTTP middlewares run as a stack around every request, listed outermost first. `response_timing` is the outermost, `refresh_db_snapshot` the innermost, and a `GZipMiddleware` (responses over 512 bytes) wraps the whole stack on top:
+Six HTTP middlewares run as a stack around every request, listed outermost first. `response_timing` is the outermost and `refresh_db_snapshot` the innermost. Compression is handled by nginx in production; the Python application does not compress responses itself.
 
 | Middleware (outermost first) | Responsibility |
 |------------|----------------|