Coding Management

With Coding, administrators can create controlled API keys for coding tools. Employees then do not use their own OpenAI key, but a key managed from AI-Public.

This is especially suitable for developers, data teams, and other advanced users who want to use AI assistance in a local development environment.

Why controlled API keys?

A regular API key often provides broad access to an external AI platform. With Coding, usage goes through AI-Public. As a result, management, costs, and reporting stay with the organization.

Key benefits:

• Costs under control: set a budget per key.
• Usage traceable: usage, costs, and chats are linked to the employee.
• Central management: create, restrict, and revoke keys from the management console.
• Team-ready: manage API keys for employees and locations from a single overview.

Enabling Coding

Coding is enabled at the environment level. This can be done by an administrator with sufficient rights.

1. Go to Management Console → Coding.
2. Click on Enable Coding.
3. Read the warning about subscription costs and API costs.
4. Click on Agree and Enable.

After enabling, AI-Public processes the technical connection in the background. Administrators do not need to manage external project settings for this.

Costs

Enabling Coding may incur additional subscription costs. API costs are billed with a markup. If you have questions, contact AI-Public.

Grant access to locations

After enabling, decide which locations may use Coding.

1. Open the Management tab.
2. Select the locations where Coding may be deployed.
3. Click on Save Access.

A location administrator will only see the locations where Coding is allowed and where the administrator has rights.

Choose models

In the Management tab, choose which OpenAI models are available for Coding.

Model	Use
GPT-5.6 Luna	Fast model for everyday programming help and initial solutions
GPT-5.6 Terra	Powerful model for complex programming tasks and code analysis
GPT-5.6 Sol	Most advanced model for demanding programming work and professional use

For most users, GPT-5.6 Luna is a fast default. Choose GPT-5.6 Terra for complex repositories and GPT-5.6 Sol for tasks that require a lot of analysis, planning, or reasoning.

Compatible coding tools

Coding works with tools that can use an OpenAI-compatible endpoint. The user then needs three pieces of data:

• the Coding API key
• the AI-Public Coding endpoint URL: https://europe-west1-ai-public-pro.cloudfunctions.net/coderenOpenAiCompatibleApi/v1
• an allowed model, for example gpt-5.6-luna

AI-Public recommends OpenCode as the default tool for advanced users:

Tool	Recommended for	Note
OpenCode	Developers and advanced users	Agentic coding tool for terminal, desktop, and IDE. Powerful, but requires experience with code, git, and local projects.
Aider	Users who explicitly want pair programming in the terminal	Technically compatible, but not the recommended default tool for Public environments.

Other tools such as Continue, Cline, Roo Code, and Sourcegraph Cody Enterprise can be technically useful if they support their own OpenAI-compatible base URL.

Not every coding tool is automatically suitable

A tool is only suitable if you can set up your own OpenAI-compatible base URL. Tools like Claude Code and Codex CLI are therefore not suitable for Coding.

Reporting

The Reporting tab shows usage and costs for the chosen period. By default, the last 7 days are shown.

The reporting includes:

• total number of requests
• total costs
• number of tokens
• breakdown by location
• breakdown by user

Disabling Coding

You can temporarily disable Coding via the toolbar button.

When Coding is disabled:

• existing API keys cannot be used
• reporting and historical data remain preserved
• Coding can be re-enabled later

Disabling does not remove existing reporting.