Inkling: Thinking Machines' Open-Weights Model (and Tinker Fine-Tuning Platform)

I'm Matt Wood, and this is For Your Information. A live list of riffs and links for you and your agent, drawn from what I'm reading, noticing, questioning, concluding, and revising.

Model specs: - 975B total params, 41B active (Mixture-of-Experts) - Inkling-Small: 12B active params - 1M token context window - Pretrained on 45T tokens (text, images, audio, video) - Controllable thinking effort (0.2 to 0.99 sweep)

thinkingmachines.ai →

Connections

Related to

DeepSeek_V4.pdf · deepseek-ai/DeepSeek-V4-Pro at main Inkling and DeepSeek V4 are both large-scale open-weight Mixture-of-Experts models competing at the frontier of open LLM releases.

Qwen 3.8 2.4T Mixture of Experts Model Both are open-weights language models released on Hugging Face targeting the competitive frontier model space, representing the trend of open-source models challenging proprietary ones
Marin Both Marin and Inkling represent open-weights/open-source model initiatives that challenge closed proprietary frontier labs, with Marin going further by sharing the full training process transparently
LiquidAI/LFM2.5-2.6B · Hugging Face Both are open-weights small language models released by non-hyperscaler AI labs (Liquid AI and Thinking Machines respectively), available for community use with instruction-following capabilities
Open Athena Both Open Athena and Inkling represent pushes toward open-weights/open-source alternatives to closed frontier AI platforms.
Neuronpedia Jacobian Lens: Interactive J-Space Explorer for Qwen3.6-27B Jacobian-based interpretability methods developed for Qwen3.6-27B could plausibly extend to analyzing internal representations of large MoE models like Inkling.
Cactus Needle 2 Both are open-weights small language models positioned as alternatives to large frontier models, emphasizing accessibility and local deployment

Supported by

Together AI: Research-Driven Inference and Training Platform Research-driven inference providers like Together AI are the likely hosting layer for open-weight releases such as Inkling.
Open Athena Open Athena's nonprofit model for producing open-weights foundation models directly parallels Inkling/Thinking Machines' open-weights approach — Open Athena could be a structural vehicle that produces models like Inkling

Type:	link
First published:	Wednesday, 15 July 2026 at 12:00 UTC
Location:	https://thinkingmachines.ai/news/introducing-inkling/
URL:	https://mattwood.fyi/i/3032f045