How to Use Cline, Kilo Code and Continue with Bumo Router
To connect Cline, Kilo Code or Continue to Bumo Router, use https://router.bumo.ai/v1, your Bumo API key and a model ID from the international gateway's live catalogue. Cline calls the provider OpenAI Compatible, Kilo adds it as a Custom provider, and Continue uses provider: openai with apiBase in YAML.
You only need to configure the editor you already use. The workflow has three steps: get a key, enter your connection settings, then check a small request before running a coding task.
This guide checks the public catalogue and official configuration documentation as of October 21, 2026. It is a configuration reference, not a claim that every model has passed an end-to-end coding-agent test in all three extensions. Bumo Router is our API product.
1. Get the key and choose an available model
Open Bumo Router, sign in and use API Key to create your key. Check your balance under Finance before making paid requests. The international site bills in USD; its recharge implementation uses card payment through Stripe. We did not complete a payment for this guide, so check the options and final amount shown in your checkout.
Keep these three values together:
| Connection detail | Value |
|---|---|
| Base URL | https://router.bumo.ai/v1 |
| API key | Your own Bumo Router international API key |
| Example model ID | gemini-3.8-flash |
The live international model list returned these example IDs during our check:
| Model family | Exact ID to copy |
|---|---|
| Gemini | gemini-3.8-flash |
| Claude | claude-opus-5-5 |
| GPT | gpt-6.1-sol |
These are catalogue entries, not a speed or coding-quality ranking. Copy the exact id from Bumo's list when switching models. The IDs above have no vendor prefix; do not add one copied from another gateway. Review the current model prices before selecting a model for a long task.
The same key can connect all three clients. If you later use them regularly, separate keys make it easier to revoke access to one client. Store keys locally and keep them out of repositories and screenshots.
2. Fill in your editor's settings
All three examples below use the gateway's OpenAI-compatible Chat Completions interface. The provider setting describes the request format; it does not restrict the model to the OpenAI family.
Cline
Open Cline's settings using the gear icon and enter:
| Cline field | Value |
|---|---|
| API Provider | OpenAI Compatible |
| Base URL | https://router.bumo.ai/v1 |
| API Key | Your Bumo international key |
| Model / Model ID | gemini-3.8-flash |
Save the configuration and confirm this model is selected for the mode you are using. If your installation has separate settings for different modes, check the active one before testing. Advanced context and output limits should match the selected model; a larger number in the editor does not increase the API's actual limits.
Field names and the compatible-provider selection follow Cline's official configuration guide.
Kilo Code
In the VS Code extension, open Settings → Providers → Custom provider. Fill the dialog as follows:
| Kilo field | Value |
|---|---|
| Provider ID | bumo-intl |
| Display name | Bumo Router |
| Provider API | OpenAI Compatible |
| Base URL | https://router.bumo.ai/v1 |
| API key | Your Bumo international key |
| Models | Select or add gemini-3.8-flash |
bumo-intl is your local connection name, not the API model ID. Kilo can fetch models from a compatible models endpoint. If discovery fails, enter the exact model ID manually. Submit the dialog, then select the added model in the model picker.
These fields follow Kilo's current OpenAI-compatible provider documentation. Older extension versions may present the provider directly in a dropdown. Keep the same base URL, key and model ID. For this recipe, keep the Chat Completions format selected when changing between the listed model families.
Continue
Open your local Continue YAML configuration and add this model entry. If you already have a models list, merge the entry into it instead of creating a second top-level list.
| Continue YAML field | Value |
|---|---|
provider | openai |
apiBase | https://router.bumo.ai/v1 |
apiKey | Your Bumo international key |
model | gemini-3.8-flash |
roles | chat for the initial check |
name: Bumo Router Local
version: 1.0.0
schema: v1
models:
- name: Bumo Gemini
provider: openai
model: gemini-3.8-flash
apiBase: https://router.bumo.ai/v1
apiKey: YOUR_BUMO_API_KEY
roles:
- chat
Replace the placeholder locally, save, and select Bumo Gemini in Continue. This config adds chat; it does not configure embeddings, autocomplete or codebase indexing. Those features need their own suitable models and settings. The apiBase mapping follows Continue's compatible-provider documentation.
3. Check a small request, then check agent behavior
Start with a short prompt such as: “Explain JavaScript Array.map in one sentence.” A successful reply confirms more than opening the public model list, which does not prove that your key or balance is valid.
Next, provide a small, non-sensitive function and ask the model to explain it without making changes. For agent use, test a single edit in a disposable project, inspect the proposed diff and run its test. Basic chat success alone does not establish tool-calling compatibility.
Finally, compare the request with Bumo's usage records. An agent task can send multiple requests while reading files, calling tools and retrying. The editor's estimate and the gateway's billed usage may therefore differ. Keep the task small until both its behavior and cost are clear.
Fix the connection at the layer that failed
| Symptom | First check |
|---|---|
| Configuration will not load | YAML indentation, duplicate models keys, or an unchanged placeholder |
| Authentication error | Key belongs to the international gateway, is complete and has not been revoked |
| Model not found | Copy the exact ID from router.bumo.ai/v1/models |
| Wrong endpoint or 404 | Use the /v1 base URL above; do not append /chat/completions in these recipes |
| Balance or key-limit error | Inspect the account balance, key restrictions and actual error detail |
| Chat works but agent tools fail | Check the chosen model's tool support and the client's mode and request format |
Avoid changing the key, model and URL simultaneously while debugging. Start from the three values in the first table, then change one setting at a time.
Looking for DeepSeek, Kimi, GLM or Qwen?
Those families were absent from Bumo's international model list at the time of this check. They do appear in the China gateway's catalogue, but that is a different endpoint with RMB billing and a WeChat payment flow. Do not assume an international balance or key transfers to it.
If that payment setup fits you, review the China gateway. Otherwise, compare the specific model's availability and payment options on OpenRouter. Our Bumo Router vs OpenRouter comparison explains the catalogue and billing differences. Neither the international signup nor the examples above promise access to the China-only catalogue.
Ready to connect your existing editor? Open Bumo Router and get an API key, select a listed model and finish the short-request check before starting a larger task.