Platform hostname

The Qubrid platform now lives at https://qubrid.com instead of https://platform.qubrid.com. That includes the dashboard, playground, API keys, inference logs, and the inference API (/api and /v1). Documentation remains at https://docs.platform.qubrid.com. Existing API keys, model IDs, and request paths are unchanged. The previous host continues to work during the transition. See the Platform Hostname Migration guide to update bookmarks and integrations.

Model Comparison

Model Comparison lets users evaluate Qubrid serverless models side by side before choosing one for production. With Model Comparison, users can:

• Compare models with each other in a single view
• Select a minimum of 2 models and a maximum of 4 models per comparison
• Review streaming responses, latency, and token metrics across selected models
• Run the same prompt against multiple models to judge quality and performance
• Make faster model selection decisions using direct head-to-head results

API Cost Estimator

API Cost Estimator helps users forecast and plan their model usage expenses before deploying or adding credits. With the cost estimator, users can:

• Select any model to simulate expected usage costs
• Input estimated daily input and output token usage
• View real-time cost projections for daily and monthly usage
• Understand base pricing for selected models (input/output token rates)
• Get smart recommendations for credit top-ups based on usage patterns
• Make informed decisions on model selection and scaling before deployment

New Dashboard

Dashboard is a centralized control panel designed to give users a complete overview of their AI usage, resources, and platform activity in one place. With the new Dashboard, users can:

• Monitor available credits and usage at a glance
• Discover and access trending or latest models from the Models Hub
• Track key metrics such as total tokens, active VMs, and deployments
• Perform quick actions like accessing templates or opening the model catalog
• Manage API keys and interact with Serverless APIs بسهولة
• Explore GPU resources and available configurations
• View recent activity including model usage, inference, and deployments
• Access live demos, documentation, and support directly from the dashboard

Serverless Model Catalog

The serverless model catalog makes it easy to discover and call hosted models without managing GPUs. With the catalog, users can:

• Browse available chat and code models in one place
• Compare pricing and model IDs before calling the API
• Authenticate with a Qubrid API key
• Use the OpenAI-compatible /v1 endpoints from their own apps and tools
• Scale usage with pay-per-token billing
• Move from model selection to production inference quickly

Model Playground

Playground is a unified interface that allows users to test and experiment with all models and APIs from a single page. With Playground, users can:

• Evaluate model performance using custom prompts
• View key metrics such as TTFT, TPS, and total tokens
• Adjust model parameters to fine-tune outputs
• Generate API keys and copy API requests directly
• Configure system prompts and include few-shot examples
• Enable or disable streaming directly from the UI