This release upgrades the default AI model to GPT-5.2 with its 1M token context window, enabling larger translation batches, fewer API calls, and new reasoning capabilities.
✨ Highlights
- GPT-5.2 as Default Model - 4x larger context budget (50K vs 12K tokens)
- Reasoning Effort Control - New low/medium/high setting for GPT-5 models
- Automatic API Adaptation - Seamlessly handles GPT-5 vs GPT-4 API differences
- MCP Server Feature - New website section with documentation link
🆕 Added
- GPT-5.2 Support
- 1M token context window for larger translation batches
- reasoning_effort parameter (low/medium/high) for GPT-5 models
- Automatic API adaptation:
- max_completion_tokens (GPT-5) vs max_tokens (GPT-4)
- json_schema structured output (GPT-5) vs json_object (GPT-4)
- developer role (GPT-5) vs system role (GPT-4)
- Model Capabilities Module (modelCapabilities.ts)
- isGPT5Model(), isReasoningModel() utility functions
- getModelCapabilities(), buildTokenParams(), buildReasoningParams()
- Reasoning Effort UI in AI Settings dialog (GPT-5 only)
- Persisted to localStorage as dqm_reasoning_effort
- Full i18n support (EN/DE/ES)
- MCP Server feature card on website with documentation link
🔄 Changed
- Default model: gpt-4.1-mini → gpt-5.2
- Context budget: 12,000 → 50,000 tokens
- Website hero badge: "GPT-5.2 Powered AI"
- Website MSW: Refactored to useMSW() hook for better lifecycle control
📚 Documentation
- Updated AI-FEATURES.md with GPT-5.2 capabilities and cost estimates
- Updated API-REFERENCE.md with new default model
- Updated EXAMPLES.md and TROUBLESHOOTING.md
- Wiki regenerated from docs/
📦 Installation
npm install @crownpeak/dqm-react-component@1.3.0💡 Migration
localStorage.setItem('dqm_openai_model', 'gpt-5.2');No breaking changes. The new GPT-5.2 default is backward compatible. To use the previous model: