Export
The link always opens the latest version of this page.
Ask an AI about it
Open in ChatGPTA new chat that reads this entry from the site Open in ClaudeThe same, in ClaudeShare the link instead of a file, and they will always see the latest version.
Files are a snapshot of today. The live page has anything newer.
Inkling
Open-weights Mixture-of-Experts transformer model with 975B total parameters (41B active), supporting 1M token context window, trained on 45 trillion tokens of multimodal data
- Category
- foundation model for tinkering 1 source
Inkling is an open-weights Mixture-of-Experts transformer language model designed for multimodal processing.
Architecture and Scale
The model features 975 billion total parameters with a Mixture-of-Experts design that activates 41 billion parameters during inference. It supports a 1 million token context window and was trained on 45 trillion tokens of multimodal data.
Capabilities
Inkling operates across multiple modalities, supporting text, image, audio, and video inputs1. The model is optimized to run on specialized chips clusters.
Versions
Version 1.0 represents both the affected and latest version1.
Facts 2
| Attribute | Value | Confidence | Sources | Cited |
|---|---|---|---|---|
| Vendor | Thinking Machines Lab | low | 1 | 2 |
| Category | foundation model for tinkering | low | 1 | source ↗ |
Version 1
Platform 1
Neighbourhood
Developed by 3
Official links
Frequently mentioned together 16
Sources 13
- Inkling: Our Open-Weights Model · Hacker News (YC)
- Alibaba's Qwen3.8 re-educated with a more Western worldview · The Register: Security
- European AI flag bearer Mistral's new open weights model is 'Le Chonk' · The Register: Security
- The current balance of power in open models · Hacker News (YC)
- video PlayBook: A Programmable Paper Notebook · Lobsters
5 of 13 sources shown · Pro unlocks the full list