Tidet Desktop
Connect Tidet Desktop to your Dedicated Gateway and run it on IG1 AI sovereign models.
Prerequisites
Before you start, make sure you have:
- Tidet Desktop installed — Download the latest version from app.tidet.ai/download
- Your Dedicated Gateway URL — e.g.
https://api.xxx.ig1.ai, wherexxxis your organization trigram - An API key — Generated from your Dedicated Gateway admin console or the User Interface
You need a valid Tidet account (Free or any tier plan) to use the Tidet Desktop application.
Which setup applies to you?
How Tidet Desktop gets its models depends on your Tidet plan. Identify your case in the table below, then follow the matching instructions.
| Your Tidet plan | Who configures the models? | What do you do? |
|---|---|---|
| Free or Cloud (individual use) | You, directly in the Tidet Desktop app | Follow Step 1A: Individual setup below |
| Business or Enterprise — team member | Your organization's admin does it once for everyone | Nothing. Models are pushed to your app automatically. Go straight to Step 2: Verify your configuration |
| Business or Enterprise — organization admin | You, once in Tidet Cloud, for the whole organization | Follow Step 1B: Organization setup |
Step 1A: Individual setup (Free and Cloud tiers)
Configure the models yourself, inside the Tidet Desktop app.
1. Add the IG1 AI provider
- Open the Command Center (
Cmd+Shift+P/Ctrl+Shift+P, or click the search icon in the title bar) - Go to Models & agents
- Select an existing profile, or create a new one
- Under LLM Provider, click + New Configuration
- As the provider type, select OpenAI Compatible
Then fill in the provider details:
| Field | Value |
|---|---|
| Provider Name | IG1 AI (or any name you prefer) |
| Base URL | Your Dedicated Gateway URL, e.g. https://api.xxx.ig1.ai |
| API Key | Your Dedicated Gateway API key |
Tidet talks to IG1 AI through the OpenAI-compatible API format, which the Dedicated Gateway fully supports.
2. Assign the model slots
Once the provider is saved, assign models to the three slots below.
Main Model
The default model, used for chat, reasoning, and agent tasks.
- Under Main Model, select your IG1 AI provider
- In the Model ID field, enter one of the following:
| Model ID | Best For |
|---|---|
glm53-flash-agent | Deep debugging, algorithm design, complex system modeling |
glm53-flash-agent-low | Deep debugging, algorithm design, complex system modeling — with minimal reasoning |
qwen35-122b-a10b-thinking-general | Deep reasoning tasks |
qwen35-122b-a10b-thinking-coding | Pair programming for algorithms, refactoring, and system design |
qwen36-35b-a3b-thinking-general | Lightweight reasoning tasks |
qwen36-35b-a3b-thinking-coding | Quick coding assistance and simple refactoring |
coding-max-thinking | Autonomous agents with preserved reasoning (virtual model) |
coding-max | Deep debugging, algorithm design, complex system modeling (virtual model) |
coding-pro-thinking | Reasoning tasks (virtual model) |
- Tick the Vision option — all IG1 AI models support vision (text, code, and OCR)
Autocomplete Model
Used for inline code completions as you type.
- Under Autocomplete Model, select your IG1 AI provider
- Enter a fast instruct model, such as:
general-maxgeneral-progeneral-stdglm53-flash-chat-lowqwen35-122b-a10b-instruct-generalqwen35-122b-a10b-instruct-creativeqwen36-35b-a3b-instruct
- Tick the Vision option
If you don't need autocomplete, leave this slot empty or disable autocomplete in the Tidet settings.
Light Model
A lighter, faster model used for quick tasks and background operations.
- Under Light Model, select your IG1 AI provider
- Enter a lightweight model, such as:
general-progeneral-stdqwen35-122b-a10b-instruct-generalqwen35-122b-a10b-instruct-creativeqwen36-35b-a3b-instruct
- Tick the Vision option
3. Save
Click Save. Continue to Step 2: Verify your configuration.
Step 1B: Organization setup (Business and Enterprise tiers)
This section is for organization admins only. Team members don't configure anything — models are applied to their desktop apps automatically.
On the Business and Enterprise tiers, the model provider is configured once at organization level in Tidet Cloud, instead of on each user's machine.
1. Add the IG1 AI provider for the organization
- Open Model & Agents in the Configuration section of Tidet Cloud
- Create a new profile for the organization (or edit the existing one)
- Under LLM Provider, click + New Configuration
- As the provider type, select OpenAI Compatible
Then fill in the provider details:
| Field | Value |
|---|---|
| Provider Name | IG1 AI (or any name you prefer) |
| Base URL | Your Dedicated Gateway URL, e.g. https://api.xxx.ig1.ai |
| API Key | Your organization's API key |
2. Assign the model slots
Assign the Main, Autocomplete, and Light model slots exactly as described in Step 1A — the available models and recommendations are the same for both setups.
3. Save
Click Save. The configuration is applied to all members of the organization automatically. Team members can check that it works with Step 2.
Step 2: Verify your configuration
- Close the Command Center (
Esc) - Start a new chat (
Cmd+N/Ctrl+N) - Check the model indicator in the chat header — it should show your selected IG1 AI model
- Send a test message, for example:
Hello! Can you confirm which model you are running on?
You should receive a response from the IG1 AI model. Tidet is now running on IG1 AI models through your Dedicated Gateway.
Available IG1 AI Models
LLM Models
| Model ID | Best For |
|---|---|
coding-max-thinking | Autonomous agents with preserved reasoning (virtual model) |
coding-max | Deep debugging, algorithm design, complex system modeling (virtual model) |
coding-pro-thinking | Reasoning tasks (virtual model) |
general-max | High-velocity coding, code review, architectural decisions (virtual model) |
general-pro | Fast, accurate, and ready for any mixed workload (virtual model) |
glm53-flash-chat-low | High-velocity coding, code review, architectural decisions — with minimal reasoning |
glm53-flash-chat | High-velocity coding, code review, architectural decisions — with configurable reasoning |
glm53-flash-agent-low | Deep debugging, algorithm design, complex system modeling — with minimal reasoning |
glm53-flash-agent | Deep debugging, algorithm design, complex system modeling — with configurable reasoning |
qwen35-122b-a10b-instruct-general | Fast, accurate, and ready for any mixed workload |
qwen35-122b-a10b-instruct-creative | Copywriting, storytelling, and brainstorming |
qwen35-122b-a10b-thinking-general | Deep reasoning tasks |
qwen35-122b-a10b-thinking-coding | Pair programming for algorithms, refactoring, and system design |
qwen36-35b-a3b-thinking-general | Lightweight reasoning tasks |
qwen36-35b-a3b-thinking-coding | Quick coding assistance and simple refactoring |
qwen36-35b-a3b-instruct | Fast, cost-effective general-purpose tasks |
Recommended Configuration
| Slot | Recommended Model ID |
|---|---|
| Main Model | coding-max-thinking |
| Autocomplete Model | general-pro |
| Light Model | general-max |
Troubleshooting
Models Not Appearing
- Verify your Base URL is correct and accessible
- Check that your API key is valid and has not expired
- Ensure model IDs match exactly (they are case-sensitive)
Connection Errors
- Confirm your Dedicated Gateway URL follows the pattern
https://api.xxx.ig1.ai - Check that your network can reach the gateway (try opening the URL in a browser)
- Verify your API key has permissions for the models you are trying to use
Slow Responses
- Switch the Light Model or Autocomplete Model to
general-profor faster responses - Reserve
coding-max-thinkingfor complex coding and reasoning tasks only