v5.5.6 - Llama 4 Maverick Support

What's New

Changed

• Groq default model updated to Llama 4 Maverick (meta-llama/llama-4-maverick-17b-128e-instruct)
  • 400B total parameters (17B active with 128 experts)
  • 128K token context window
  • Multimodal vision support (up to 5 images)
  • Production-ready on Groq with ultra-fast inference

Upgrade

pip install --upgrade llmswap