[Feature]: Add support for Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber

1. Why Do You NEED This Feature?

I NEED this feature because:

• Google recently released the Gemini 3.6 Flash, 3.5 Flash-Lite, and 3.5 Flash Cyber models on July 21, 2026.
• Gemini 3.6 Flash is Google's latest workhorse model that offers frontier-level intelligence optimized for multi-step orchestration, full-stack code refactoring, and general reasoning at a much lower cost ($1.50 / $7.50 per 1M tokens) and higher speed.
• Currently, users cannot leverage these newly launched Gemini models in Forge, limiting their productivity compared to using other platforms.

2. What Is NOT Possible Right Now?

Currently, I cannot:

• Configure or select gemini-3.6-flash, gemini-3.5-flash-lite, or gemini-3.5-flash-cyber in Forge configuration files or interactive menus.
• Use these high-performance, cost-efficient models for multi-step code generation or multi-agent workflows within the Forge terminal.

3. What WILL Be Possible With This Feature?

With this feature, I will be able to:

• Use gemini-3.6-flash, gemini-3.5-flash-lite, and gemini-3.5-flash-cyber natively in Forge.
• Run multi-step agentic workflows with 1M context window at extremely low latency and cost.
• Select these models interactively via :model command.

Proposed Solution (User Experience)

From a user's perspective, the feature should work like this:

• The new models gemini-3.6-flash, gemini-3.5-flash-lite, and gemini-3.5-flash-cyber will be available under google_ai_studio and vertex_ai providers.
• When running :model, users can select these models from the list.

Alternatives Considered

None. These are official Google Gemini model releases and must be supported natively.

Feature Category

AI/LLM Integration

Priority/Impact

High - Would significantly improve my workflow

Examples from Other Tools

Other major development tools and SDKs have already added native support for Gemini 3.6 Flash.

Pre-submission Checklist

• I have searched existing issues and confirmed this is not a duplicate
• I am willing to submit a PR to implement this feature