● Covered by 4 sources · 4 reportsMedium impact
Inkling: Thinking Machines Releases Open-Weights Multimodal AI Model
Key points
- Inkling has 975 billion total parameters and 41 billion active.
- It supports a context window of up to 1 million tokens.
- Trained on 45 trillion tokens covering text, image, audio, and video.
- Open-weight model allows custom modifications and enterprise use.
- Mixture-of-Experts design enhances efficiency and reduces costs.
Overview of Inkling Release
Thinking Machines Lab has unveiled its new multimodal language model, Inkling. With approximately 975 billion total parameters and 41 billion active, it supports text, audio, and images through an open-weight structure, allowing developers and enterprises to freely modify and tailor the model to their needs.
Key Specifications
Inkling features a large 1 million token context window and employs a highly efficient Mixture-of-Experts design, which selectively utilizes active parameters to manage computational demands and maintain performance. The model was trained on a vast dataset of 45 trillion tokens, encompassing text, images, audio, and video formats.
Enterprise Applications
The open-source nature of Inkling makes it attractive for businesses that need customizable AI solutions that can be hosted on-premises or within private cloud environments. Its resistance to censorship is an added benefit, offering more reliable and sensitive topic handling.
Impact and Industry Perspectives
Inkling represents a significant shift from traditional one-size-fits-all AI models. By providing open weights and vast capacity for customization, Thinking Machines is aiming to empower enterprises and developers to innovate and implement adaptable AI solutions across various domains and applications.
Conclusion
Inkling’s release signals a new phase in AI development focused on flexibility and domain-specific adaptation. This model could potentially pave the way for new multimodal reasoning applications, offering businesses a competitive edge in leveraging AI innovatively.
✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →
The daily brief
One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.
One email a day. Unsubscribe in one click, any time.
Today's brief
Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.
▶ Play today's brief
New every morning, and the back catalogue is archived by date.
Primary sources
GitHub huggingface/blog GitHub huggingface/OpenEnv GitHub huggingface/trl GitHub thinking-machines-lab/tinker-cookbook GitHub radixark/miles GitHub lightseekorg/tokenspeedHow outlets covered it
Thinking Machines has open-sourced Inkling, a multimodal language model designed for enterprise use. It offers a balance between cost and performance while providing resistance to censorship, making it appealing for businesses requiring custom AI solutions.
Inkling, a new AI model with 975B parameters and multimodal capabilities, has been released with open weights, enabling customization. It supports a context window of up to 1M tokens and aims to balance performance with flexible application across various domains.
Thinking Machines Lab released Inkling, an open-weight AI model that allows external developers to modify it. Targeted at enterprises, it features a mixture-of-experts design for efficiency and is positioned as an adaptable alternative to traditional large AI models.
Inkling, a multimodal LLM with approximately 1 trillion parameters, has been launched by Thinking Machines on Hugging Face. This model, which supports various input modalities, including text, audio, and images, is designed for advanced reasoning and domain adaptation, making it a significant development in the AI field.