Changelog

NovaAI 3.0 is our largest platform update to date, introducing new AI capabilities, significant performance improvements, and a refreshed developer experience. This release focuses on helping developers build intelligent applications faster with improved APIs, better SDKs, and more powerful automation tools. Alongside new features, we've invested heavily in documentation, platform stability, and overall reliability to make day-to-day development smoother and more productive.

Introducing the Agents API

NovaAI now includes a dedicated Agents API for building intelligent assistants capable of completing multi-step tasks. Agents can reason through user requests, retrieve information from external knowledge sources, invoke tools through Function Calling, and return structured responses without requiring developers to manually orchestrate every interaction.

The new API follows the same design principles as the rest of the NovaAI platform, allowing developers to integrate autonomous workflows using familiar SDKs and authentication methods.

Faster Chat Models & Streaming

We've significantly improved the performance of our Chat Completion models. Response latency has been reduced across supported regions, while streaming reliability has been enhanced to deliver smoother real-time responses for chatbots, coding assistants, and interactive AI applications.

Developers should also notice more consistent output quality during long conversations thanks to improvements in context handling and message processing.

Improved Developer Experience

This release includes updates across all official SDKs with simplified client initialization, improved error handling, and better TypeScript support. We've also redesigned large parts of our documentation to include clearer examples, practical guides, expanded API references, and an AI-powered documentation assistant that helps developers find answers faster.

Whether you're integrating NovaAI for the first time or maintaining an existing application, these improvements reduce setup time and make everyday development more efficient.

Improvements

• Improved response latency across Chat Completion models.

• Faster streaming performance with smoother token delivery.

• Enhanced Vision and Image Generation quality.

• Better SDK consistency across JavaScript, Python, Go, and PHP.

• Expanded API examples and developer documentation.

• Improved error messages for authentication and validation failures.

• Faster Vector Store indexing and retrieval performance.

• Optimized infrastructure for higher request throughput.

We're incredibly excited to share NovaAI 3.0 with the developer community. As always, your feedback continues to shape the platform, and we'll keep shipping improvements focused on performance, reliability, and developer experience in upcoming releases.

This maintenance release focuses on improving the stability, security, and overall reliability of the NovaAI platform. While this update doesn't introduce new APIs or developer-facing features, it includes numerous optimizations behind the scenes to deliver faster response times, stronger request validation, and a more dependable development experience. These improvements help ensure your applications remain secure, responsive, and ready for production workloads.

Faster & More Reliable Platform

We've optimized request processing across our infrastructure to reduce latency and improve consistency during periods of high traffic. Streaming connections are now more stable, API responses are more predictable under load, and several backend optimizations have improved overall platform performance without requiring any changes to existing integrations.

Enhanced Security

This release strengthens authentication and request validation across the platform. Additional security checks have been introduced to better protect API requests, improve credential verification, and reduce the risk of unauthorized access. We've also enhanced monitoring systems to detect unusual activity more effectively while maintaining compatibility with existing applications.

Fixes & Improvements

• Improved API response times across core services.

• Enhanced authentication and request validation.

• Fixed intermittent streaming connection issues.

• Improved SDK stability during long-running requests.

• Optimized infrastructure for higher request throughput.

• Resolved inconsistencies in API error responses.

• Strengthened platform monitoring and security checks.

• Updated internal dependencies for improved reliability.

We're excited to introduce the new Vision and Audio APIs, expanding NovaAI beyond text generation into multimodal AI. Developers can now analyze images, transcribe speech, generate natural-sounding audio, and build richer AI experiences using a unified API and the same familiar SDKs. This release makes it easier to create applications that understand both visual and spoken content while maintaining a consistent developer experience across the platform.

Introducing the Vision API

The new Vision API enables models to interpret images alongside text, making it possible to build applications for image captioning, document understanding, screenshot analysis, visual question answering, and object recognition. Simply provide an image with an optional prompt, and the model returns context-aware insights in a structured format.

Introducing the Audio API

The Audio API combines speech-to-text and text-to-speech capabilities into a single interface. Developers can transcribe recordings, generate realistic voice responses, detect spoken languages, and build voice-enabled assistants without relying on multiple services or APIs.

Improvements

• Added multimodal support for image and audio processing.

• Introduced new Vision and Audio endpoints across all official SDKs.

• Improved image analysis accuracy for documents and screenshots.

• Enhanced speech recognition with better language detection.

• Added natural voice presets for text-to-speech generation.

• Included new API examples and integration guides in the documentation.

• Optimized processing performance for faster image and audio requests.

• Expanded SDK support with consistent request and response formats.

Today we're releasing JavaScript SDK 2.4, focused on improving the overall developer experience for JavaScript and TypeScript applications. This update simplifies client initialization, improves SDK consistency, and introduces several quality-of-life enhancements that make integrating NovaAI faster and more reliable. Along with performance optimizations, we've refined documentation, expanded examples, and improved error reporting to help developers build with greater confidence.

Improved Developer Experience

The SDK now offers a cleaner initialization process, more consistent method naming, and enhanced TypeScript definitions for better autocomplete and type safety. We've also updated code examples throughout the documentation to reflect the latest SDK patterns, making it easier for both new and existing developers to integrate NovaAI into their projects.

Better Performance & Reliability

Version 2.4 includes improvements to request handling, streaming performance, and overall SDK stability. Error messages are now more descriptive, making debugging easier, while internal optimizations reduce overhead during repeated API requests and improve reliability for long-running applications.

Improvements

• Simplified SDK initialization with a cleaner client configuration.

• Improved TypeScript support and autocomplete definitions.

• Enhanced streaming performance for real-time responses.

• Added clearer API error messages and validation feedback.

• Optimized request handling for better reliability.

• Updated code samples across the documentation.

• Improved compatibility with modern JavaScript runtimes.

• Reduced SDK bundle size for faster application startup.

Bug Fixes

• Fixed an issue where streaming responses could stop unexpectedly during long-running requests.

• Resolved inconsistent TypeScript type definitions for chat completion methods.

• Fixed environment variable detection in Node.js ES Module projects.

• Corrected request retry behavior for intermittent network failures.

• Fixed duplicate error events emitted during failed API requests.

• Resolved incorrect parsing of multipart file uploads in image and audio APIs.

• Fixed race conditions affecting concurrent client instances.

• Corrected import path resolution issues in certain bundlers and build tools.