# Gemini in VS Code: Cline Custom API and Base URL Setup

> Configure Gemini in VS Code with Cline and OpenAI Compatible. Enter the Passion8 /v1 URL, API key and model, then verify Plan/Act settings and file tools.

URL: https://docs.passion8.cc/en/docs/gemini/vscode
Language: en
Publisher: Passion8
Last updated: 2026-10-11

For Gemini in VS Code, Passion8 recommends **Cline + Passion8**: the extension manages project files and tool calls, while Passion8 handles model requests. Prepare a Passion8 API key and an enabled Gemini model ID from the console.

## 1. Install and open Cline

Search for **Cline** in the VS Code extension marketplace, verify extension ID `saoudrizwan.claude-dev`, install it and open its sidebar. You can also install from the [official Cline listing](https://marketplace.visualstudio.com/items?itemName=saoudrizwan.claude-dev).

Open the project folder you intend to edit. Keep manual approval for file writes and command execution during the first task so you can review changes.

## 2. Enter the three connection settings

Open Cline settings, choose **OpenAI Compatible** as API Provider, and enter:

| Setting | Value |
| --- | --- |
| Base URL | `https://passion8.cc/v1` |
| API Key | Your Passion8 API key |
| Model ID | Copy the full enabled Gemini model ID from the console |

Save and return to the conversation. This connection uses the OpenAI compatible protocol and includes `/v1`. [Gemini CLI](https://docs.passion8.cc/en/docs/gemini) uses the native protocol at `https://passion8.cc`; do not copy its endpoint into this provider.

If you enable separate models for Plan and Act, check the provider, URL, key and model in both modes. Configuring Plan alone can leave Act using a previous provider.

## 3. Confirm requests use Passion8

Start a new task and enter:

```text
Reply only OK. Do not read files or run commands.
```

Find the matching request in the [Passion8 console](https://passion8.cc) and check its model. Next ask: “Inspect this project without editing files. Explain its startup command and test entry points.”

A text response does not prove the full coding workflow works. Check that it can read an approved file, propose an edit and display a diff. If a tool call fails, retain the error and verify your account's channel capabilities.

## 4. Configure model capabilities

If your Cline version lets you edit custom model capabilities, set context length, maximum output and image support according to the selected model and Passion8 channel. Do not copy another Gemini model's context window or enable image support based only on its name.

Complete one file read and edit using the default tool format first. Advanced tool formats, image input and cache accounting require separate verification; a model list confirms selectable models, not every capability. Keep long tasks focused and start a new conversation after completing a goal.

## Current status of official Gemini extensions and CLI

As checked on 2026-10-11, Google states that Gemini Code Assist IDE extensions and Gemini CLI login service for the free individual tier, Google AI Pro and Ultra stopped serving requests on **2026-06-18**, with migration to Antigravity. The migration announcement retains paid Gemini API key access, and the CLI configuration reference still documents a custom `GOOGLE_GEMINI_BASE_URL`.

A change to Google login does not mean Gemini models are unavailable through APIs. Passion8 setup uses a URL, key and model; Code Assist or Antigravity account login is not an installation step.

**Gemini CLI Companion** provides editor context and diff integration and needs Gemini CLI. It is not standalone Code Assist or a chat extension configured with just the three connection fields. Existing API-key CLI users can follow official [IDE integration](https://geminicli.com/docs/ide-integration/) with `/ide install` and `/ide enable`; authentication remains in the CLI configuration.

## Troubleshooting

| Symptom | Check |
| --- | --- |
| Google login opens | You installed Code Assist or selected Google's official provider; this guide uses Cline OpenAI Compatible |
| 404 | Use `https://passion8.cc/v1`; do not append `/chat/completions` or Google's `/v1beta/openai` |
| 401 / 403 | Use a Passion8 key, confirm it is active and the account can access the model |
| Plan works but Act fails | Check provider, model and key for each mode |
| Chat works but file tools fail | Check tool support, model ID and Cline tool settings |
| No matching Passion8 usage | Check for a previous provider configuration, then verify in a new task |

For other providers, see [Cline multi-model setup](https://docs.passion8.cc/en/docs/cline). For practical tasks, see [Gemini coding workflow](https://docs.passion8.cc/en/docs/gemini/workflow). To integrate the model into your own application, follow [Gemini API and SDK](https://docs.passion8.cc/en/docs/gemini/api).

## Verified sources

Checked: 2026-10-11.

- [Current Gemini Code Assist status](https://developers.google.com/gemini-code-assist/docs/set-up-gemini)
- [Gemini CLI migration announcement](https://developers.googleblog.com/an-important-update-transitioning-gemini-cli-to-antigravity-cli/)
- [Gemini CLI custom base URL](https://geminicli.com/docs/reference/configuration/)
- [Cline OpenAI Compatible](https://docs.cline.bot/provider-config/openai-compatible)