Why does WorkBuddy return 404 after adding a custom model?

The model name must exactly match the identifier shown in the model plaza. Namespaced models (e.g. deepseek/deepseek-chat) must keep their prefix.

Is my API key uploaded anywhere?

No. WorkBuddy custom model settings are stored locally in ~/.workbuddy/models.json and never leave your machine.

What base URL should I use?

Use the OpenAI-compatible endpoint https://www.apikit.cc/v1 together with your APIKIT API key.

GUIDE / WORKBUDDY × APIKIT

Use GPT, Claude & Gemini inside WorkBuddy:
the APIKIT setup guide

WorkBuddy supports custom models. Plug in your APIKIT API key and the same workspace can switch between GPT, Claude, Gemini, Grok and 50+ models on demand — about 5 minutes, no config files to edit.

⏱ ~5 minutes🧩 No coding💻 Windows / Mac🔑 One key, 50+ models

Contents Background: what APIKIT adds to WorkBuddy Step 1: Create your APIKIT account Step 2: Create an API key Step 3: Pick models in the model plaza Step 4: Add a custom model in WorkBuddy Step 5: Switch models and start chatting Which frontier model to choose Troubleshooting

00Background: what APIKIT adds to WorkBuddy

APIKIT is an AI API gateway: one account, one API key, OpenAI-compatible access to OpenAI (GPT), Anthropic (Claude), Google (Gemini), xAI (Grok) plus DeepSeek, Kimi, Qwen, GLM and more — 50+ models in total.

The APIKIT website (apikit.cc): one entry point to 50+ models

WorkBuddy, Tencent's AI workspace, officially supports custom models under Settings → Model — any OpenAI-compatible service works. Together you can switch models per task: Claude for long reports, GPT for everyday Q&A, Gemini for multimodal analysis.

🔑 Why it helps: one endpoint, one key, one balance for every model. No separate OpenAI / Anthropic / Google accounts, no overseas payment hassle.

01Step 1: Create your APIKIT account

Open www.apikit.cc/register — email or Google sign-in both work, no invitation code needed.

Open the sign-up page

Click "Get started" in the top-right corner or go to apikit.cc/register directly.

Sign in

After signing in you land in your personal console (dashboard).

💡 New accounts get a $1 free credit (applied automatically for both email and Google sign-in) — enough to validate everything before you top up.

02Step 2: Create an API key

In the console (apikit.cc/dashboard), open the "API Keys" page and create a new key.

Create the key

Click "Create key", give it a name (e.g. workbuddy), and the console generates an sk- key.

Copy it right away

The full key is shown only once at creation. Save it locally — you'll paste it into WorkBuddy next.

⚠️ Security: your key is tied to your balance. Never share it in chats or commit it to repositories.

03Step 3: Pick models in the model plaza

Open the model plaza (/model-plaza) to see every model available to your account, with exact identifiers. Copy identifiers character by character — the name you type into WorkBuddy must match exactly.

Model plaza: the small label on each card (claude-sonnet-5, gpt-5.4, …) is the identifier you'll use

The catalog updates as upstreams evolve; always trust the identifier on the card over marketing names. You'll use it in step 4.

04Step 4: Add a custom model in WorkBuddy

The only WorkBuddy-side step. Open the WorkBuddy desktop app, click your account avatar (bottom-left) → Settings → Model.

Settings → Model → Add model

In the "Custom models" block click "＋ Add model" to open the configuration dialog.

Fill in the four fields

Field	Value	Notes
Provider	`Custom`	APIKIT is not in the built-in provider list
Base URL	`https://www.apikit.cc/v1`	OpenAI-compatible endpoint; keep the trailing `/v1`
API Key	the `sk-` key from step 2	stored locally only, never uploaded
Model name	e.g. `gpt-5.4`	must match the model plaza identifier exactly, prefixes included

WorkBuddy add-model dialog (Chinese UI shown; your localization may differ): Provider = Custom, Base URL = https://www.apikit.cc/v1

Save

Click "Save". Settings persist automatically — no restart needed.

💡 Want several models?Repeat "Add model" with the same base URL and key, changing only the model name: gpt-5.4, claude-opus-5, gemini-3-flash… Add as many as you like — they all appear under the "Custom models" group in the model selector.

05Step 5: Switch models and start chatting

Back in the WorkBuddy chat, open the model selector next to the input box and pick your new model under "Custom models". Messages now run on the GPT / Claude / Gemini you chose.

The quick start page (apikit.cc/en/start) has ready-to-run Python / Node.js / cURL examples

Send a test message

Ask "which model are you?" to confirm the routing matches your selection.

Check usage in the console

The console's usage/log page shows each request's model, token consumption and cost.

🧪 Developer check (optional):verify your key with one command first:

curl https://www.apikit.cc/v1/chat/completions -H "Authorization: Bearer sk-YOUR_KEY" -H "Content-Type: application/json" -d "{\"model\":\"gpt-5.4\",\"messages\":[{\"role\":\"user\",\"content\":\"hi\"}]}"

06Which frontier model to choose

Subject to the live model plaza catalog, these are the flagship picks per family (identifiers as shown in the plaza):

Model	Family	Strengths	WorkBuddy suggestion
`gpt-5.4`	OpenAI GPT	balanced chat, writing, instruction following	default daily driverTop pick
`gpt-5.4-mini`	OpenAI GPT	fast, cost-effective	translation, summaries, batch jobs
`gpt-6-sol`	OpenAI GPT	complex reasoning, long-form planningNew gen	solution design, multi-step tasks
`claude-opus-5`	Anthropic Claude	long-form writing, code, deep analysis	reports / whitepapers, large codebasesTop pick
`claude-sonnet-5`	Anthropic Claude	speed/quality balance	everyday writing and code edits
`gemini-3.8-flash-high`	Google Gemini	multimodal, long context, fast	image understanding, very long documents
`grok-4.7`	xAI Grok	real-time topics, bold style	trend takes, brainstorming

Running the same task across models is the fastest way to find your best workflow — exactly what "one key, every model" is for.

07Troubleshooting

Symptom	Cause & fix
404 / model not found	Model name typo. Copy the identifier from the model plaza character by character; keep namespaced prefixes (e.g. `deepseek/deepseek-chat`)
401 / invalid key	Key copied incompletely or deleted. Create a new one in the console and paste the full `sk-` string
New model missing from selector	Make sure you clicked "Save" under Settings → Model; restart WorkBuddy if it still doesn't appear
Insufficient balance	Check your balance in the APIKIT console and top up; all custom models share one balance
What base URL to use	Always `https://www.apikit.cc/v1`. WorkBuddy uses the standard `/chat/completions` path — leave the "custom protocol" toggle off
Is the key safe?	WorkBuddy stores custom model settings locally (`~/.workbuddy/models.json`); keys are never uploaded. Don't share your key with anyone

Use GPT, Claude & Gemini inside WorkBuddy:the APIKIT setup guide