Doesn't know how to change models on windows (default) but knows how to do it using WSL

[SaraSSC]

Describe the bug

When asked to change models or if it knows how to do it, it doesn't know this information:" By default, Copilot utilises Claude Sonnet 4. We also support GPT-5 via an environment variable. Run COPILOT_MODEL=gpt-5 copilot to launch in GPT-5 mode. Or on Windows, run set COPILOT_MODEL=gpt-5 before running copilot." nor does it point to its repo to search for documentation.

When trying with WSL it is aware.

• new Powershell 7.5.3
• npm and node up to date

Affected version

Version 0.0.327 · Commit 0cbec74

Steps to reproduce the behavior

1. open terminal -> mine is set to use the new PowerShell
2. prompt : "which ai model is your default and it is possible to change models ? "

Outupt using PowerShell:

"Let me check the configuration help to see if there are model selection options:

✓ Check Copilot CLI configuration help
$ copilot help config
╭──────────────────────────────────────────────────────────────────────────────────────────────────╮
│ Configuration Settings: │
│ │
│ banner: frequency of showing animated banner; defaults to "once". │
│ - "always" displays it every time │
│ - "never" disables it │
│ - "once" displays it the first time │
│ │
│ beep: whether to beep when user attention is required; defaults to true. │
│ │
│ render_markdown: whether to render markdown in the terminal; defaults to true. │
│ │
│ screen_reader: whether to enable screen reader optimizations; defaults to false. │
│ │
│ theme: theme to color and stylize output; defaults to "auto". │
│ - "auto" detects terminal background and decides based on luminosity │
│ - "dark" uses contrasting ANSI colors for dark backgrounds │
│ - "light" uses contrasting ANSI colors for light backgrounds │
│ │
│ trusted_folders: list of folders where permission to read or execute files has been granted. │
│ <exited with exit code 0> │
╰──────────────────────────────────────────────────────────────────────────────────────────────────╯

● Based on my investigation, here's what I found:

Current Setup

 * My Model: Claude 3.5 Sonnet (as I mentioned earlier)
 * Your CLI Version: You have GitHub Copilot CLI installed (version 0.0.327 based on my system info)

Model Selection

From the available configuration options, I don't see explicit model selection
settings in the current GitHub Copilot CLI. The available configuration options
are:

 * Banner display
 * Beep notifications
 * Markdown rendering
 * Screen reader optimizations
 * Theme settings
 * Trusted folders

Possible Model Configuration

Model selection might be:

 1. Determined by your GitHub Copilot subscription tier
 2. Set at the organization/enterprise level if you're using enterprise Copilot
 3. Handled automatically based on the task type
 4. Available through environment variables not shown in the help

To Explore Further

You could:

 * Check your GitHub Copilot subscription settings on GitHub.com
 * Look for any .copilot configuration files in your home directory
 * Contact GitHub support if you need specific model access

The model I'm using (Claude 3.5 Sonnet) is quite capable for coding tasks, so
you're getting a high-quality AI experience regardless!"

vs on WSL output:

I'm powered by Claude 3.5 Sonnet from Anthropic - that's my underlying AI model.
This is indicated in my system information where I'm described as "GitHub
Copilot CLI, a terminal based coding agent built by GitHub" but running on
Claude's architecture.

Regarding changing models - this depends on how GitHub Copilot CLI is configured
in your specific setup:

Model Selection Possibilities

For GitHub Copilot CLI:

 * The model selection typically depends on your GitHub Copilot subscription and configuration
 * Some versions may allow model selection via command-line flags or configuration files
 * Enterprise users might have different model options available

For comparison, GitHub Copilot in VS Code:

 * Offers model selection (GPT-4, GPT-3.5-turbo, etc.)
 * Can be changed in VS Code settings
 * Different models for different use cases (chat vs completion)

How to Check Your Options

Let me help you check what's available in your current setup:

✓ Check GitHub Copilot CLI help for model options
$ gh copilot --help "

Then it suggests to use:
COPILOT_MODEL= model name

Expected behavior

Expected to know this info: By default, Copilot utilises Claude Sonnet 4. We also support GPT-5 via an environment variable. Run COPILOT_MODEL=gpt-5 copilot to launch in GPT-5 mode. Or on Windows, run set COPILOT_MODEL=gpt-5 before running copilot.

Additional context

No response

[RyanHecht]

Hi! We've added the /model slash command now to make model switching easier!

[SaraSSC]

Hi Ryan
Thank you, and keep up with the great work