Thinking Machines ships Inkling 2 Small: 594B total / 17B active open multimodal model
ariG23498 · x · 2026-11-13
Thinking Machines has released Inkling 2 Small on Hugging Face, the first member of the second generation of its open Inkling multimodal model family.
- Scale: 594B total parameters with 17B active, taking text and image inputs; it keeps the core decoder ideas of the original Inkling while redesigning almost everything else around them.
- Family overview: The first-gen flagship Inkling is a 1T parameter model natively accepting image, text, and audio inputs with a 1M context window, trained on 45 trillion tokens of text, images, audio and video; Inkling Small is a lighter 276B total / 12B active variant.
- Positioning: Built for cross-modality reasoning and domain adaptation via fine-tuning. The HF article also explains how to run Inkling 2 Small with transformers, plus inference, use cases, post-training and deployment examples.
Related event: Thinking Machines Open-Sources Inkling 2 Small Weights under Apache 2.0(26 posts)→
More from Models
- ChatGPT Dot Appears to Be Rolling Out in EU, Pro User Reports Access — kimmonismus · 2026-11-13
- Dev claims the best voice mode and computer use he's used, built on OpenAI Realtime 2 and Claude — RileyRalmuto · 2026-11-13
- Why AI Slop Is So Frustrating to Read: Hidden Subjects and Buried Actions — teortaxesTex · 2026-11-13
- MSR: Rewriting fewer than 10 dormant neurons backdoors any LLM agent, no training needed — mikeflache · 2026-11-13
- Google's Logan Kilpatrick asks users what to ship next; community wants a new Gemma — jacek2023 · 2026-11-13
- Microsoft diagnostic AI solves 83% of NEJM cases vs 36% for physicians — realmeetjames · 2026-11-13