# CodeBuddy > Connect CodeBuddy to the LMU AI API via models.json — enter the full /chat/completions endpoint to use Claude, GPT and Chinese models. URL: https://docs.lmuai.com/en/docs/tools/codebuddy CodeBuddy (Tencent Cloud's coding assistant) is an AI code editor that supports a custom model list via a `models.json` config file. This page covers connecting it to LMU AI. *** ## Install and sign in [#install-and-sign-in] 1. Download and install the build for your operating system from the official site: [https://www.codebuddy.cn](https://www.codebuddy.cn/home/) 2. Sign in to your CodeBuddy account when prompted *** ## Configure LMU AI [#configure-lmu-ai] ### Step 1: Pick the config file location [#step-1-pick-the-config-file-location] | Level | Path | Notes | | ---------------------------- | --------------------------------------- | ------------------------------------------------ | | **User level** (recommended) | `~/.codebuddy/models.json` | Global config, applies to all projects | | **Project level** | `/.codebuddy/models.json` | Project-specific, takes priority over user level | Create the file if it doesn't exist. Merge order from highest to lowest: project level → user level → built-in defaults. ### Step 2: Write the config [#step-2-write-the-config] Using the user-level `~/.codebuddy/models.json` as an example: ```json { "models": [ { "id": "claude-opus-5", "name": "Claude Opus 5 (LMU AI)", "vendor": "LMU AI", "apiKey": "sk-your-lmu-ai-api-key", "url": "https://api.lmuai.com/v1/chat/completions", "maxInputTokens": 200000, "maxOutputTokens": 32000, "supportsToolCall": true, "supportsImages": true, "supportsReasoning": true }, { "id": "gpt-5.6-sol", "name": "GPT-5.6 Sol (LMU AI)", "vendor": "LMU AI", "apiKey": "sk-your-lmu-ai-api-key", "url": "https://api.lmuai.com/v1/chat/completions", "maxInputTokens": 400000, "maxOutputTokens": 32000, "supportsToolCall": true } ] } ``` The `url` field takes the **full endpoint path**, ending in `/chat/completions`: * ✅ Correct: `https://api.lmuai.com/v1/chat/completions` * ❌ Wrong: `https://api.lmuai.com/v1` * ❌ Wrong: `https://api.lmuai.com` This is the most common reason a copy-pasted minimal example fails — plenty of tutorials treat it as a base URL, and CodeBuddy then can't reach the endpoint. The official docs state that only **OpenAI-format** APIs are supported at present, so connect to LMU AI over the OpenAI-compatible protocol. Claude models work fine too: enter a model ID like `claude-opus-5` directly, and the LMU AI backend handles the OpenAI ↔ Anthropic conversion. ### Step 3: Pick the model and start chatting [#step-3-pick-the-model-and-start-chatting] Just save the file — models.json supports **hot reload** (1-second debounce), so there's no need to restart CodeBuddy. Select your newly configured model in the chat box's model picker and you're set. Models added via models.json are automatically tagged `custom`, making them easy to spot in the UI. *** ## Field reference [#field-reference] Fields available on each entry in the `models` array: | Field | Type | Required | Notes | | ------------------- | ------- | -------- | ----------------------------------------------------------------------------------------- | | `id` | string | ✓ | Unique model identifier — use the LMU AI model ID | | `name` | string | - | Display name in the dropdown | | `vendor` | string | - | Vendor name; anything you like | | `apiKey` | string | - | The `sk-` key generated in the LMU AI console (the actual key value, not an env var name) | | `url` | string | - | Full endpoint path; for LMU AI, `https://api.lmuai.com/v1/chat/completions` | | `maxInputTokens` | number | - | Maximum input tokens | | `maxOutputTokens` | number | - | Maximum output tokens | | `supportsToolCall` | boolean | - | Whether tool calling is supported | | `supportsImages` | boolean | - | Whether image input is supported | | `supportsReasoning` | boolean | - | Whether reasoning mode is supported | *** ## Show only specific models [#show-only-specific-models] Use the top-level `availableModels` field to control which models appear in the dropdown: ```json { "models": [ { "id": "claude-opus-5", "name": "Claude Opus 5 (LMU AI)", "apiKey": "sk-your-lmu-ai-api-key", "url": "https://api.lmuai.com/v1/chat/completions", "supportsToolCall": true } ], "availableModels": ["claude-opus-5"] } ``` * Omitted or an empty array → all models are shown * Once set → only the listed model IDs appear (built-in and custom models alike) * A project-level `availableModels` **fully overrides** the user-level one rather than merging Per the official docs: after deleting the `availableModels` field, remember to remove the now-trailing `,` after the `models` array above it, or the JSON becomes invalid and the whole config stops working. *** ## Common model IDs [#common-model-ids] | Model ID | Notes | | --------------------- | ------------------------------------- | | `claude-opus-5` | Claude Opus 5 (flagship, recommended) | | `claude-sonnet-5` | Claude Sonnet 5 (balanced) | | `claude-haiku-4-5` | Claude Haiku 4.5 (fast) | | `gpt-5.6-sol` | GPT-5.6 Sol | | `glm-5.2` | GLM-5.2 | | `qwen3.8-max-preview` | Qwen 3.8 Max Preview | | `deepseek-v4-pro` | DeepSeek V4 Pro | | `kimi-k3` | Kimi K3 | Not sure which model to enter? Go to the [Model Gallery](/en/docs/guide/models) to see every available model ID and copy it with one click. For a 1M context window, write the model ID with the `[1M]` suffix (e.g. `claude-opus-5[1M]`) and raise `maxInputTokens` accordingly. *** ## Troubleshooting [#troubleshooting] ### Config not taking effect? [#config-not-taking-effect] 1. Check that the **JSON is valid** (a missing or trailing comma is the usual culprit) 2. Confirm the **file path** is right (`~/.codebuddy/models.json`) 3. Confirm every entry has the required `id` field 4. Confirm the file was actually **saved to disk** (hot reload has a 1-second debounce) ### Model missing from the dropdown? [#model-missing-from-the-dropdown] * If `availableModels` is set, check that the model ID is listed there * Check that the entry in the `models` array is complete ### Getting a 401 / 404? [#getting-a-401--404] * **401**: is `apiKey` the `sk-` key generated in the LMU AI console? * **404**: `url` is almost certainly missing `/chat/completions` — it must be the full path *** ## Notes [#notes] * `url` must be the **full endpoint path** `https://api.lmuai.com/v1/chat/completions`, not just `/v1` or the bare domain * CodeBuddy supports **only the OpenAI format**; Claude models go through LMU AI's protocol conversion * `apiKey` takes the actual key value — env var names aren't supported * The config **hot reloads**, so saving is enough; no restart needed * Each model entry needs its own `apiKey` and `url` * Project-level config overrides user level per `id`, while `availableModels` is replaced wholesale rather than merged