January 2025 Updates

Here's everything we shipped in January 2025.

🤖 New Models

OpenAI

GPT-4o-mini - Smaller, faster, cheaper version of GPT-4o

- Input: $0.15 / 1M tokens

- Output: $0.60 / 1M tokens

- 128K context window

Anthropic

Claude 3.5 Sonnet - Latest Claude model with improved reasoning

- Input: $3.00 / 1M tokens

- Output: $15.00 / 1M tokens

- 200K context window

- Enhanced code generation capabilities

Google

Gemini 2.0 Flash - Ultra-fast multimodal model

- Input: $0.075 / 1M tokens

- Output: $0.30 / 1M tokens

- Native image understanding

DeepSeek

DeepSeek-V3 - Open-source model with GPT-4 level performance

- Input: $0.27 / 1M tokens

- Output: $1.10 / 1M tokens

- Best cost-performance ratio

🚀 Performance Improvements

API Latency: Reduced average latency by 23% across all models
Streaming: Improved first-token latency for streaming responses
Connection Pooling: Better connection reuse reducing overhead

🛠️ Platform Updates

Dashboard

New real-time usage charts
Model performance comparison tool
Export usage data to CSV

API

Added x-request-id header for request tracing
Improved error messages with actionable suggestions
New /v1/models endpoint with filtering options

SDKs

TypeScript SDK v2.1.0 - Better type inference
Python SDK v2.1.0 - Async support improvements
Go SDK v1.0.0 - Initial release!

🐛 Bug Fixes

Fixed rate limit calculation for batch requests
Resolved timeout issues with long-running requests
Fixed JSON parsing for certain Unicode characters
Corrected token counting for non-English text

📅 Coming Soon

Fine-tuning support for select models
Team management features
Advanced analytics dashboard
Webhook notifications

Feedback

Have feedback or feature requests? We'd love to hear from you:

Email: feedback@cozhub.ai
Twitter: @cozhub

Thank you for building with COZHUB! 💜

January 2025 Changelog