Caspio 68.0

Extensions are plug-and-play modules that connect your Caspio account to third-party services, enabling various functionalities.

AI-Powered GPT Connect

In this release, we introduced the following enhancements: 

• Persistent instructions and reference files – You can define default instructions and attach reference files, such as policies, schemas, documentation, or examples, that AI-Powered GPT Connect uses with every request. This lets you guide how the AI behaves and ensure responses are consistently based on your own material, without retraining the model or reattaching files each time.
• Multi-file support for prompts – You can now include multiple files in a single AI-Powered GPT Connect prompt instead of being limited to one. This enables more advanced requests, such as analyzing, comparing, or combining information across documents, and allows the AI to produce more complete and useful responses based on all relevant files at once.
• Configurable Top-P parameter – You can now control how narrowly or broadly AI-Powered GPT Connect selects its responses using the Top-P setting, alongside Temperature. Lower Top-P values make responses more focused and predictable, while higher values allow more variation and creativity. This gives you finer control over AI behavior, helping you match responses to the task, whether you need strict, consistent outputs or more flexible, exploratory ones.
• New models and reasoning-effort controls – Added support for GPT-5.1 and GPT-5.2, with expanded per-model reasoning-effort settings. Reasoning effort controls how much internal compute the model allocates to multi-step thinking and deliberation: higher settings produce deeper, more reliable reasoning for complex tasks, while lower settings favor faster responses at lower cost. Because higher reasoning effort consumes more compute per request, it directly increases cost, allowing you to trade off depth and accuracy against performance and spend.

Learn more about AI-Powered GPT Connect.