Inkling‑Small, released by Thinking Machines, offers 276 B total parameters with 12 B active weights and supports a 1 M token context window. The model is available in NVFP4 format on Hugging Face (thinkingmachines/Inkling‑Small‑NVFP4) and as GGUF binaries from Unsloth (unsloth/Inkling‑Small‑GGUF). A detailed blog post and additional resources are provided by the authors.