Key Info

Hugging Face announced day-one support for Thinking Machines Lab's Inkling 2 Mini, an A3B/26B model, across transformers, llama.cpp and MLX.

Highlights

  • Inkling 2 Mini uses an A3B/26B configuration, described as a practical size for local setups
  • Support landed simultaneously in transformers, llama.cpp and MLX
  • Hugging Face staff said they had extensively vibe-tested it as a coding companion
  • A blog post accompanies the release