What's New

Changelog

Stay up to date with the latest features, improvements, and fixes

8
Releases
35
New Features
25
Improvements
7
Bug Fixes
Stay Updated
Get notified when we release new features
RSS Feed
v2.1.0
Minor
January 20, 2025

DeepSeek R1 & Enhanced Analytics

Added DeepSeek R1 reasoning model and redesigned usage analytics dashboard.

New Features
  • •DeepSeek R1 reasoning model now available - state-of-the-art reasoning capabilities
  • •New usage analytics dashboard with real-time charts and breakdowns
  • •Model comparison tool - compare up to 4 models side by side
  • •Prompt template library with 50+ ready-to-use templates
  • •Webhook notifications for usage alerts and API events
Improvements
  • •Response streaming now 40% faster with optimized chunking
  • •Dashboard load time reduced by 60% with lazy loading
  • •Better error messages with actionable suggestions
  • •Enhanced rate limit headers with remaining quota info
Bug Fixes
  • •Fixed token counting for multi-byte characters
  • •Resolved timeout issues with long-running requests
  • •Fixed API key rotation not taking effect immediately
v2.0.0
Major
January 10, 2025

COZHUB 2.0 - Major Platform Update

Complete platform redesign with new features, improved performance, and expanded model support.

New Features
  • •Completely redesigned dashboard with modern UI
  • •Added 20+ new AI models including GPT-4o, Claude 3.5, Gemini 2.0
  • •New playground with side-by-side model comparison
  • •Team collaboration features with role-based access control
  • •API versioning support (v1, v2) for backwards compatibility
  • •New SDK packages for TypeScript, Python, Go, and Ruby
Improvements
  • •40% faster response times with edge-optimized routing
  • •Rebuilt infrastructure on global CDN for lower latency
  • •Enhanced security with SOC 2 Type II compliance
  • •Improved documentation with interactive examples
Deprecated
  • •Legacy v0 API endpoints (migration guide available)
  • •Old dashboard UI (auto-redirects to new dashboard)
v1.9.0
Minor
December 15, 2024

Gemini 2.0 Flash & Image Generation

Added Google Gemini 2.0 Flash model and expanded image generation capabilities.

New Features
  • •Google Gemini 2.0 Flash - fastest multimodal model
  • •FLUX 1.1 Pro image generation model
  • •Batch API for processing multiple requests efficiently
  • •Usage export to CSV/JSON for billing integration
Improvements
  • •Image generation API now supports custom aspect ratios
  • •Better handling of concurrent requests
  • •Improved model availability monitoring
Bug Fixes
  • •Fixed memory leak in streaming connections
  • •Resolved billing discrepancy for cached responses
v1.8.0
Minor
November 20, 2024

Claude 3.5 Sonnet & Function Calling

Added Anthropic Claude 3.5 Sonnet and improved function calling support.

New Features
  • •Anthropic Claude 3.5 Sonnet - best-in-class for coding tasks
  • •Enhanced function calling with parallel tool execution
  • •New /models endpoint for dynamic model discovery
  • •Request logging with 30-day retention
Improvements
  • •Function calling response time reduced by 25%
  • •Better JSON mode reliability across all models
  • •Enhanced streaming for tool use responses
Security
  • •Added API key scoping for read-only and write access
  • •Implemented request signing for enhanced security
v1.7.0
Minor
October 25, 2024

Embeddings API & Vector Search

Launched embeddings API with support for multiple embedding models.

New Features
  • •Embeddings API with OpenAI text-embedding-3 models
  • •Cohere embed-v3 multilingual embeddings
  • •Jina embeddings for long document support
  • •Batch embedding endpoint for bulk processing
Improvements
  • •Reduced embedding latency by 30%
  • •Added dimension reduction option for embeddings
  • •Better documentation for RAG use cases
v1.6.0
Minor
September 30, 2024

OpenAI o1 Models & Reasoning

Added OpenAI o1-preview and o1-mini reasoning models.

New Features
  • •OpenAI o1-preview - advanced reasoning model
  • •OpenAI o1-mini - cost-effective reasoning
  • •New reasoning token tracking in usage stats
  • •Extended context support up to 128K tokens
Improvements
  • •Better handling of long-running reasoning requests
  • •Improved timeout handling for complex queries
Bug Fixes
  • •Fixed context truncation for very long conversations
  • •Resolved rate limit calculation for streaming requests
v1.5.0
Minor
August 15, 2024

Vision API & Multi-modal Support

Added vision capabilities for image understanding across multiple models.

New Features
  • •Vision API support for GPT-4V and Claude 3 Vision
  • •Image URL and base64 input support
  • •Multi-image conversations
  • •Image token estimation for cost prediction
Improvements
  • •Unified vision API across all providers
  • •Better image compression for faster uploads
  • •Enhanced error messages for unsupported formats
v1.4.0
Minor
July 10, 2024

Llama 3.1 & Open Source Models

Expanded support for open source models including Llama 3.1 family.

New Features
  • •Meta Llama 3.1 8B, 70B, and 405B models
  • •Mistral Large and Mistral Nemo
  • •Qwen 2 series from Alibaba
  • •New model filtering by license type
Improvements
  • •Open source models now with 99.9% availability
  • •Reduced pricing for open source model inference
  • •Better system prompt handling for Llama models