> Source: Onevium official documentation > Article: Connect providers and choose models > Original URL: https://onevium.com/docs/providers > Language: English > Updated: 2026-09-09 > Applies to: 1.1.22 > Feature status: Released --- # Connect providers and choose models Compare Onevium provider presets, distinguish Claude Code sign-in from API keys, and verify a connection with text and tool requests. ## Choose a connection method Onevium has two paths for accessing models: | Method | When to use it | How it works | | ------------------- | ------------------------------------------------------------------------------- | ------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Claude Code sign-in | Use an existing Claude Code account login or subscription authorization | Open the Claude Code connection status panel and use its sign-in action. Onevium opens the bundled Claude Code login flow in its terminal. After signing in, select a model under **Claude Code** in the model menu | | Provider API key | Use the Anthropic API, another preset vendor, or your team's compatible gateway | Open **Settings → Providers**, add the credential, and select that provider in the model menu | Signing in to Onevium, activating a Onevium license, and obtaining model access are separate steps. A Claude subscription login does not create an API key for the provider form. ## Prepare the right account and credential Have the API key for the correct product, region, and account ready. Confirm that the account can use your intended model. Coding Plans, general API platforms, and regional accounts may have different endpoints; choose the preset that matches your account's product. Enter secrets in the **API Key** field, not in chat or screenshots. For Bearer authentication, enter the raw credential without adding a `Bearer ` prefix. ## Review the built-in presets Version 1.1.22 includes these **nine chat presets**. They use Anthropic/Claude-compatible interfaces. An ordinary OpenAI-compatible endpoint is not a direct substitute. | Display name | Region or product | Prefilled Base URL | Authentication | | ----------------- | -------------------- | ---------------------------------------- | --------------------- | | Anthropic | Official API | `https://api.anthropic.com` | API key / `x-api-key` | | DeepSeek | General API platform | `https://api.deepseek.com/anthropic` | Bearer token | | GLM (CN) | China GLM Code Plan | `https://open.bigmodel.cn/api/anthropic` | Bearer token | | z.ai (Global) | Global GLM Code Plan | `https://api.z.ai/api/anthropic` | Bearer token | | Kimi Coding Plan | Kimi Coding Plan | `https://api.kimi.com/coding` | API key / `x-api-key` | | Moonshot (CN) | Moonshot China API | `https://api.moonshot.cn/anthropic` | API key / `x-api-key` | | Moonshot (Global) | Moonshot global API | `https://api.moonshot.ai/anthropic` | API key / `x-api-key` | | MiniMax (CN) | China Code Plan | `https://api.minimaxi.com/anthropic` | Bearer token | | MiniMax (Global) | Global Code Plan | `https://api.minimax.io/anthropic` | Bearer token | Presets supply the name, endpoint, authentication method, and model configuration. You normally only need to enter an API key. The model catalog is supplied by the app; it is not a live query of the models available to your account. Verify availability with an actual request. **Image generation is configured separately:** | Display name | Purpose | Preset endpoint | Credential | | --------------------- | --------------------------------------------- | -------------------------------------------------- | -------------- | | Google Gemini (Image) | Generate images through Google's official API | `https://generativelanguage.googleapis.com/v1beta` | Google API key | This preset appears under **Media Providers**, not in the chat model groups. It is not a Gemini chat connection. Changing its endpoint does not establish support for third-party image proxies. ## Example: connect the DeepSeek preset 1. Open **Settings → Providers** and find **Add Provider**. 2. Find **DeepSeek** in the chat provider list and click **Connect**. 3. Paste your DeepSeek account's API key. The preset supplies the Anthropic-compatible endpoint and Bearer authentication shown above. 4. Review **Visible models**. Show or hide models as needed, keeping at least one visible. Leave advanced options at their defaults initially. 5. Click **Connect**. When the entry appears under **Connected Providers**, close Settings. 6. Open a conversation. In the composer's model menu, find the DeepSeek group you just added and select a model your account can use. Saving an entry does not test the upstream connection or switch every conversation. Follow the same process for other chat presets, using the credential for the corresponding account. ![DeepSeek preset connection form](https://onevium.com/assets/docs/providers-connect-en.png "Original components with isolated example data. Enter your own API key and choose visible models; saving still requires verification with a real request.") ## Select models, switch connections, and manage visibility Check the provider and model in the composer before starting a task. The model menu groups entries by provider; identically named models can belong to different connections. Use the interface's selection as evidence, not the model's own description of itself in a reply. Open **Settings → Providers → your connected entry → Models** to show or hide models. This controls the menu, not the validity of the API key. Keep at least one model visible for each provider. To add a real model identifier, see [custom providers](https://onevium.com/docs/providers/custom#model-routing). Provider connections are shared within the app, while conversations retain their own selections. The composer also remembers your recent choice. Channels and automations have separate configurations; changing the current conversation's model does not change all of them. **Disconnect deletes the provider entry; it does not pause it.** First check conversations, channels, and automations that still reference the entry and move them to a working connection. For an ordinary model switch, use the composer without deleting the old provider. ## Verify configuration, then make real requests First click the stethoscope icon beside the connected provider to open **Provider Doctor**. It checks local credential configuration, authentication conflicts, the endpoint, and model records. **It does not make an upstream request.** A passing report still needs real verification. Start with a small text-only request: ```text Reply with one sentence confirming you received this message. Do not read files or run commands. ``` After the selected connection replies, verify a tool you actually need. Open a test project containing README.md and send: ```text Read only README.md in the project root and summarize its run instructions. If it does not exist, say so. Do not create or modify files or run commands. ``` Check for an actual file-reading tool call and its result. A successful text response does not prove image, long-context, or universal tool support. Verify tasks intended for a channel or automation through that entry point as well. ## Troubleshoot a connection | Symptom | Check first | | -------------------------------------------------------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | 401 or authentication failure | Does the key match the selected region and product? Does the authentication field match `x-api-key` or Bearer? Avoid adding a second set of authentication environment variables | | Model not found or access denied | Check the selected provider, real model ID, and account permissions. Catalog entries do not mean your account has access | | A custom connection still lists Claude models | Model mappings or custom model entries may be missing. Follow the [custom configuration steps](https://onevium.com/docs/providers/custom) | | Timeout or connection failure | Check Base URL, network access, and **Settings → Network proxy**. Do not put a network proxy address in Base URL. A passing proxy test does not validate model credentials | | Text works but tools fail | Inspect the failing tool and upstream error. Confirm Claude-compatible tool support and test a model with the required capability | | The saved key cannot be decrypted after moving computers | Edit the provider, re-enter the original account's API key, and save | | A channel or automation still uses the old model | Inspect the provider and model selection in that channel or automation's own configuration | ## Next steps For a team gateway or manually configured fields, read [Add a custom provider](https://onevium.com/docs/providers/custom). Once the connection works, complete [your first task](https://onevium.com/docs/getting-started) and review [permissions](https://onevium.com/docs/permissions) and [settings and data](https://onevium.com/docs/settings-and-data).