IQNECT uses OpenAI’s API as its core AI engine. As such, IQNOX does not directly train or version foundational models. Updates to the AI models are managed and deployed by OpenAI and may include improvements to performance, safety, capabilities, or new model releases (e.g., GPT‑4, GPT‑4o).
AI Model Update Cadence #
OpenAI periodically updates its models with new capabilities and performance enhancements. These updates occur independently of IQNOX’s release schedule and may involve:
- Upgrading default models (e.g., from GPT-3.5 to GPT-4)
- Modifying safety filters or response behavior
- Introducing new features such as retrieval-augmented generation (RAG) or vision/multimodal support
IQNOX monitors OpenAI’s model announcements, API logs, and changelogs to track these updates.
Change Monitoring and Customer Impact Review #
When a significant upstream AI model change is identified, IQNOX performs:
- Impact assessments to evaluate how the new behavior might affect output quality, consistency, or integration behavior
- Regression testing across commonly used workflows
- Prompt structure validation to ensure stability of context and expected results
Customer Notification Process #
If a model update introduces a material change that could affect accuracy, performance, or customer workflow, IQNOX will:
- Provide proactive notification to affected customers
- Offer guidance or configuration options (e.g., model pinning or prompt tuning) where possible
- Document changes in our release notes or support portal (iqnect.ai/support)