Open WebUI 0.12.4: our test log

Open WebUI 0.12.4 installed with uv in 11.6 s (6,977 MB in the virtualenv), was healthy about 11 to 12 s after start, and answered a chat through its API in 0.68 s using a local Ollama model.

Result at a glance

What we found

• Chat through Open WebUI's API returned "42" with HTTP 200 in 0.68 s; its own usage block reported 24 prompt tokens and 3 completion tokens.
• The installed environment is very large (6,977 MB) ; plan disk accordingly. We did not inspect which dependencies account for it.
• Startup printed a CORS warning (allow origin * by default) and an insecure-key-length warning from the JWT library for our short test secret; both are configuration points to fix in a real deployment.

What we ran, step by step

# Install with uv on Python 3.12
uv venv -p 3.12 work/py/open-webui && VIRTUAL_ENV=work/py/open-webui uv pip install --no-cache -q open-webui

# Start the server
DATA_DIR=work/owui-data3 OLLAMA_BASE_URL=http://127.0.0.1:11434 open-webui serve --port <port>

# Chat request
POST /api/chat/completions with model qwen2.5-3b:latest

Raw output (excerpts)

{"id":"qwen2.5-3b:latest-2e7c387b-abc5-4dbd-93da-8e6cf26c1eca","created":1792988710,"model":"qwen2.5-3b:latest","choices":[{"index":0,"logprobs":null,"finish_reason":"stop","message":{"role":"assistant","content":"42"}}],"object":"chat.completion","usage":{"input_tokens":24,"output_tokens":3,"total_tokens":27,"prompt_tokens":24,"completion_tokens":3,"response_token/s":10.93,"prompt_token/s":68.47,"total_duration":631495510,"load_duration":2663423,"prompt_eval_count":24,"prompt_eval_duration":350540000,"eval_count":3,"eval_duration":274475000,"approximate_total":"0h0m0s","completion_tokens_details":{"reasoning_tokens":0,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}}}
HTTP 200 in 0.682456s

Gotchas and things that broke

• The first API call returned "Not authenticated" until we created a local admin user on a fresh data directory and used a bearer token; the browser UI sign-up flow was not exercised.
• Model id must match what Ollama lists (here qwen2.5-3b:latest).

What we did not test

• The browser UI, RAG over documents, web search, multi-user roles, and Docker deployment.
• One request; a smoke test, not a load test.

Reproduce it

Versions change. These commands install whatever is latest on the day you run them unless a version is pinned, so expect different results later. The version we measured is in the table above.

uv venv -p 3.12 .venv && VIRTUAL_ENV=.venv uv pip install open-webui
DATA_DIR=./data OLLAMA_BASE_URL=http://127.0.0.1:11434 .venv/bin/open-webui serve --port <port>

How to read this test

Every number on this page was measured on this one machine, which also ran other jobs, so timings are indicative and vary run to run. A pass means our small task worked once or as stated, a fail means it did not work as we ran it; neither says how the tool behaves with a vendor's frontier model, on a GPU, or at scale. Where a tool could not run without an account or a GPU, we say so instead of guessing.

See the testing rig for the hardware, the local model and the rules we follow, and the test hub for every other test.

Related tests: workflow automation

• Flowise: Flowise failed to install from npm on Node 24.22.1 after 106.2 s: the better-sqlite3 native build needed a C++ toolchain this machine lacks.
• Langflow: Langflow 1.13.0 installed in 19.76 s (1.7 GB) and its backend was healthy 91.5 s after launch, serving 26 starter flows and 416 components.
• n8n: n8n 2.44.6 installed from npm on Node 24 in 1 min 54 s (2,343 packages), imported a three-node workflow from the CLI and executed it: a Code node returned 42 and a request to a local model returned "42".

All test logs · How we test

FAQ

Sources

• Open WebUI repository (fetched 2026-10-25)

Related

• All test logs
• Methodology and testing rig
• Study: install size and first-task results across 16 frameworks