2.0.3 · Red Carpet

What's New

Models updates

This release embraces models precision: improved parameter configuration to match the model vendors, a new native xAI-flavored responses API, a new model vendor, and all the newest models including the already legendary Kimi 2.5, great for code.

• Kimi K2.5 with image understanding and reasoning
• Gemini 3 Flash with 4-level thinking (high, medium, low, minimal)
• GPT 5.2 with x-high reasoning and native search
• GPT Image 1.5 for image generation
• Novita.ai full provider support with automatic capability detection
• LMStudio model capability support and images
• xAI Responses native protocol support with X post search and code execution

Also New

• Google Drive integration - attach docs, sheets, images with optimal LLM conversion
• Inworld speech synthesis with streaming playback
• Model duplication for custom configurations
• AI Injector, collapsible model groups, PDF auto-OCR
• Copy as-seen, reorder messages in cleanup mode
• Anthropic and OpenAI API fixes, security patches, CJK/IME input

Full Changelog: https://github.com/enricoros/big-AGI/compare/v2.0.2...v2.0.3

Get Started

Available now at big-agi.com, via Docker, or self-host from source.