Key Info

ThinkyMachines' Inkling 2 Mini is now supported in Ollama 0.40.17 (pre-release), a model small enough to run locally on Macs with 32GB+ memory.

Highlights

  • Each token loops through its middle layers three times, giving deeper reasoning at just 5.4B active parameters.
  • Targeted at coding, agentic workflows such as computer use, design with strong visual reasoning, and finance tasks.
  • Run locally with ollama run inkling-2-mini, or in Pi via ollama launch pi --model inkling-2-mini.
  • Pre-release install: curl -fsSL ... | OLLAMA_VERSION=0.40.17-rc3 sh.
  • Early user reports it runs well on a 32GB Mac.