Connect Open WebUI to Plan2API

Open WebUI settings for Plan2API

Open WebUI talks to Plan2API through a standard OpenAI API connection, so a single connection brings in Claude, Codex, Gemini and Kimi models from the subscriptions you've hooked up, the same way you'd use a Claude plan as an API in any OpenAI-compatible tool. Here are the exact values to drop into your Open WebUI setup:

Add the Plan2API connection

1. Sign up with your Google account, connect at least one subscription (signing in with Claude, entering a one-time device code for ChatGPT Codex, signing in with the Google account that has Google AI Pro or Ultra, or pasting a Kimi Code API key), and copy the p2a_ API key you get after connecting.
2. In Open WebUI, open Settings > Admin > Connections, find Manage OpenAI API Connections, and click + Add Connection.
3. Set Connection Type to External, put https://api.plan2api.com/v1 in the URL field, and paste your p2a_ key into the API Key field so it goes out in an Authorization: Bearer header.
4. Leave the API Type toggle on Chat Completions (the default), because requests go to POST /v1/chat/completions and Plan2API has no /v1/responses endpoint.
5. Click Verify Connection beside the URL field, wait for Server connection verified, then click Save.

Choose which models show up

If you leave Model IDs empty under Advanced, Open WebUI calls GET /v1/models and loads every model Plan2API serves, even for providers you haven't connected yet. Pick a model from an unconnected provider and the chat returns HTTP 403 with no_credentials, so fill Model IDs with only the model ids from the subscriptions you've actually connected to keep your model picker clean (once Model IDs is filled, Open WebUI stops calling /models).

You can also set a Prefix ID like p2a under Advanced if you run several OpenAI-compatible connections side by side and want to tell them apart. Open WebUI shows names like p2a.claude/claude-sonnet-5 in the picker and strips the p2a. prefix before sending the request out, whether you're calling Claude, Codex, Kimi or Gemini from a Google AI plan.

Here are the main model ids and prefixes you can add to Model IDs:

Run Open WebUI in Docker with Plan2API

When you start the container, pass OPENAI_API_BASE_URL set to https://api.plan2api.com/v1 and OPENAI_API_KEY set to your p2a_ key. If you're wiring up more than one OpenAI-compatible endpoint at once, use OPENAI_API_BASE_URLS and OPENAI_API_KEYS instead, with the values separated by semicolons.

Keep in mind that ENABLE_PERSISTENT_CONFIG defaults to True, which means after the container's first start the settings saved in Open WebUI's database win over environment variables and any later edits have to go through Settings > Admin > Connections (unless you set ENABLE_PERSISTENT_CONFIG=False, where environment variables win and UI edits reset on restart). Also, AIOHTTP_CLIENT_TIMEOUT is unset (no timeout) by default, which suits long reasoning replies, while AIOHTTP_CLIENT_TIMEOUT_MODEL_LIST defaults to 10 seconds.

This command adds OPENAI_API_BASE_URL and your key to Open WebUI's standard Docker quick start command and serves the app at http://localhost:3000:

docker run -d -p 3000:8080 \
  --add-host=host.docker.internal:host-gateway \
  -v open-webui:/app/backend/data \
  -e WEBUI_SECRET_KEY=your-secret-key \
  -e OPENAI_API_BASE_URL=https://api.plan2api.com/v1 \
  -e OPENAI_API_KEY=p2a_your_key \
  --name open-webui --restart always \
  ghcr.io/open-webui/open-webui:main

Generate images on Pro and Scale

1. In Open WebUI, go to Settings > Admin > Images.
2. Set Image Generation Engine to Default (Open AI).
3. Enter https://api.plan2api.com/v1 as the API Base URL and your p2a_ key as the API key.
4. Set the model to gpt-image-2 (for model names starting with gpt-image, Open WebUI leaves out response_format and reads the b64_json PNG that comes back) and pick an image size of auto, 1024x1024, 1536x1024 or 1024x1536.
5. If you configure via Docker environment variables instead, set ENABLE_IMAGE_GENERATION=true and IMAGE_GENERATION_MODEL=gpt-image-2; IMAGES_OPENAI_API_BASE_URL and IMAGES_OPENAI_API_KEY default to OPENAI_API_BASE_URL and OPENAI_API_KEY from your chat connection, while IMAGE_GENERATION_ENGINE defaults to openai and IMAGE_SIZE defaults to 512x512.

Image generation requires a Pro or Scale plan (the Free plan returns HTTP 403 with type plan_limit on POST /v1/images/generations and POST /v1/images/edits), runs on your own connected ChatGPT (Codex) subscription, and takes 15 to 45 seconds per render or up to about 110 seconds at high quality.

Direct Connections for per-user keys

If you share one Open WebUI instance across a team or household and don't want everyone burning through a single shared key, an admin can turn on Direct Connections (which Open WebUI marks experimental) in Settings > Admin > Connections > Direct Connections or by setting ENABLE_DIRECT_CONNECTIONS=true. Once that's on, each user goes to User Settings > Connections and puts in https://api.plan2api.com/v1 and their own p2a_ key, and the browser calls Plan2API straight from the client while keeping the key in the browser's local storage instead of on the Open WebUI server, so every person runs on their own connected subscriptions and plan.

Request limits and Plan2API plans

Every chat message you send in Open WebUI is one chat request, and because Open WebUI turns on title, tag and follow-up generation by default (ENABLE_TITLE_GENERATION, ENABLE_TAGS_GENERATION and ENABLE_FOLLOW_UP_GENERATION all default to True), those background tasks fire extra calls against the task model so a single message in the UI can spend several requests. If you're on the Free plan, either switch those toggles off in Settings > Interface or set TASK_MODEL_EXTERNAL to a model you don't mind spending requests on.

Plan2API Free is $0 (no card needed) and gives you 1 connected subscription, 50 chat requests a day (resetting at 00:00 UTC, with x-plan2api-requests-remaining on responses and HTTP 429 daily_requests on the 51st), 1 request per second with bursts up to 20, and 4 requests at once. Pro is $12 a month or $99 a year with no daily request cap, 3 subscriptions, 10 requests per second (bursts up to 600), 16 requests at once and image generation. Scale is $29 a month or $249 a year with 25 subscriptions and 32 requests at once, as listed on the Plan2API pricing page. Plan2API never bills per token, though your provider's own subscription usage limits still apply, and if you connect more than one account from the same provider, a request that hits a limit moves to the next account before any output is sent (or returns 429 subscription_quota_exhausted with Retry-After if all of them are out, as covered in the API error codes documentation).