Test Builder — Enhanced Model Selection and Advanced Generation Engine

September 15, 2025

Two major upgrades ship for the Test Builder in this release. First, the model selector has been completely rebuilt with a three-tier interface supporting 54+ models across OpenAI, Anthropic Claude, xAI Grok, and Google Gemini — with real-time cost estimates before you generate.

Second, the prompt generation engine has been upgraded to a modular, multi-strategy architecture with dedicated systems for complexity analysis, format variation, and intelligent strategy selection. Together, these changes give you far more control and transparency over what model you use, what it costs, and how your prompts are constructed.

New Features

• Enhanced model selector with 54+ models — The Test Builder now features a three-tier selection interface: choose your AI provider first, then a category (fast & affordable, balanced, premium, or specialized), then the specific model. Each model displays its context window size, capabilities, and recommendations to help you choose the right tool for the job.
• Real-time cost estimation — Before you generate, the model selector shows estimated per-request costs broken down by input and output tokens. You can see exactly what each model will cost before committing, making it easy to balance quality against budget.
• Multi-provider support — Generate prompts using models from OpenAI, Anthropic Claude, xAI Grok, and Google Gemini, all from a single interface. Switch between providers without leaving the builder.
• Advanced prompt generation engine — The generation backend has been rebuilt as a modular system with dedicated components for complexity analysis, format variations, prompt construction, and strategy selection. This produces higher-quality prompts that are better adapted to your specific complexity/output combination.
• Format variation system — The engine now supports multiple output format variations, letting it tailor the generated prompt’s structure to different format types rather than using a one-size-fits-all approach.

Improvements

• Model metadata and categorization — Every available model now includes detailed metadata: context window size, capability tags, cost data, and categorization (fast & affordable, balanced, premium, specialized). This makes it much easier to compare models at a glance.
• Cost tracking integration — All test generation operations now feed into the enhanced cost monitoring system, so you can track spending across models and optimize your usage over time.
• Visual consistency — The enhanced model selector matches the Test Builder’s existing design language with consistent borders, gradients, and responsive behavior. Dropdown menus now have solid backgrounds (no more transparency issues) and the container expands properly when additional options appear.