From ace84b4ae961e0d0507ad61c5a63bb6ac7a729cc Mon Sep 17 00:00:00 2001
From: Classic298 <27028174+Classic298@users.noreply.github.com>
Date: Fri, 31 Jul 2026 23:26:38 +0200
Subject: [PATCH] perf: use the orjson codec for outbound Ollama request bodies
 (#27811)

`routers/ollama.py` serializes the outbound body with stdlib `json` on six inference paths: `/api/chat`, the OpenAI-compatible completions and chat completions proxies, embeddings, the Anthropic messages proxy, and responses. All six carry a full conversation or an embedding batch.

They now go through `JSONCodec`, which selects orjson when `ENABLE_ORJSON` is set. Every one is passed to `send_request`, which hands it to aiohttp as `data=`; aiohttp encodes `str` as UTF-8 and derives `Content-Length` from the encoded bytes. None is hashed, cached, length-measured or persisted.

Admin model management keeps stdlib: `/api/unload`, `/api/pull`, `/api/delete` and `/api/show` serialize fixed one- or two-key dicts, as do the blob download and upload progress events and the error frames. Codec dispatch on those costs about what it saves.

With `ENABLE_ORJSON` unset, which is the default, `JSONCodec` is stdlib `json` and this call site behaves exactly as before.
---
 backend/open_webui/routers/ollama.py | 12 ++++++------
 1 file changed, 6 insertions(+), 6 deletions(-)

diff --git a/backend/open_webui/routers/ollama.py b/backend/open_webui/routers/ollama.py
index b91aed2ac4..53575ab449 100644
--- a/backend/open_webui/routers/ollama.py
+++ b/backend/open_webui/routers/ollama.py
@@ -1149,7 +1149,7 @@ async def generate_chat_completion(
 
     return await send_request(
         f'{url}/api/chat',
-        payload=json.dumps(payload),
+        payload=JSONCodec.dumps(payload),
         key=get_api_key(url_idx, url, api_configs),
         user=user,
         stream=form_data.stream,
@@ -1244,7 +1244,7 @@ async def generate_openai_completion(
 
     return await send_request(
         f'{url}/v1/completions',
-        payload=json.dumps(payload),
+        payload=JSONCodec.dumps(payload),
         key=get_api_key(url_idx, url, api_configs),
         user=user,
         stream=payload.get('stream', False),
@@ -1295,7 +1295,7 @@ async def generate_openai_embeddings(
 
     return await send_request(
         f'{url}/v1/embeddings',
-        payload=json.dumps(payload),
+        payload=JSONCodec.dumps(payload),
         key=get_api_key(url_idx, url, (await Config.get('ollama.api_configs', {}))),
         user=user,
         metadata=metadata,
@@ -1352,7 +1352,7 @@ async def generate_openai_chat_completion(
 
     return await send_request(
         f'{url}/v1/chat/completions',
-        payload=json.dumps(payload),
+        payload=JSONCodec.dumps(payload),
         key=get_api_key(url_idx, url, api_configs),
         user=user,
         stream=payload.get('stream', False),
@@ -1404,7 +1404,7 @@ async def generate_anthropic_messages(
 
     return await send_request(
         f'{url}/v1/messages',
-        payload=json.dumps(payload),
+        payload=JSONCodec.dumps(payload),
         key=get_api_key(url_idx, url, api_configs),
         user=user,
         stream=payload.get('stream', False),
@@ -1462,7 +1462,7 @@ async def generate_responses(
 
     return await send_request(
         f'{url}/v1/responses',
-        payload=json.dumps(payload),
+        payload=JSONCodec.dumps(payload),
         key=get_api_key(url_idx, url, api_configs),
         user=user,
         stream=payload.get('stream', False),