TL;DR
Thinking Machines has publicly released the full weights of its new AI model, Inkling, under an open license, challenging the industry norm. The model is not the most powerful but offers transparency and flexibility for users.
Thinking Machines has released the full weights of its latest AI model, Inkling, under an open license on Hugging Face. This marks a notable departure from industry practice, as the model is openly available but not the most powerful or closed-model options on the market. The release emphasizes transparency and ownership, making Inkling accessible for independent testing and deployment, which is significant for the AI community and potential enterprise users.
Inkling is a 975-billion-parameter mixture-of-experts transformer supporting multimodal inputs—text, images, and audio—with a 1-million-token context window. It was trained on 45 trillion tokens of diverse data, including text, images, audio, and video, and features a unique encoder-free design for multimodal processing. The model’s weights are openly available under Apache 2.0 license on Hugging Face, enabling users to download, modify, and deploy independently.
Thinking Machines also introduced a smaller variant, Inkling-Small, with 276 billion total parameters and 12 billion active, which reportedly matches or exceeds the larger model on several benchmarks thanks to an improved pre-training approach. The training details include hybrid optimization on NVIDIA systems and over 30 million reinforcement learning rollouts, with some training data generated by open-weight models like Kimi K2.5.
However, the company clarified that open weights are not equivalent to open source; the training data and pipeline are not publicly disclosed. Additionally, reports suggest that Thinking Machines maintains a separate Model Acceptable Use Policy (AUP) restricting certain uses, such as surveillance and deception, which raises questions about the model’s openness and restrictions beyond the Apache license.
Implications of Open-Weight Release for AI Development
The release of Inkling’s full weights under an open license represents a shift toward greater transparency and user ownership in AI development. It allows organizations and researchers to inspect, fine-tune, and deploy the model independently, reducing reliance on proprietary APIs and closed systems. This move could influence industry standards, encouraging more companies to adopt open-weight policies and challenge the dominance of closed models. However, the potential restrictions via the separate AUP may complicate the narrative of true openness, raising questions about how freely the model can be used and modified.
Building Intelligent Applications with Spring AI: Develop Practical Java Solutions with Generative AI, Multimodal Models, and Agents
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Industry Norms and the Significance of Open Weights
Traditionally, most large AI models are released with limited access, often via APIs or with proprietary weights, to protect commercial interests and control usage. While some open models exist, they are often accompanied by restrictions or lack transparency regarding training data and pipelines. The recent trend toward releasing full model weights openly is relatively new but growing, driven by calls for transparency, reproducibility, and democratization of AI research. Notably, the industry has seen debates around open-source versus proprietary models, especially concerning safety, misuse, and commercial viability.
Thinking Machines’ approach with Inkling—open weights under Apache 2.0 but with a potential usage policy—fits into this evolving landscape, emphasizing openness but also raising questions about the scope of restrictions and true transparency.
“Our goal is to empower the community with access to powerful models while maintaining responsible use through our policies.”
— Thinking Machines spokesperson
AI model training and fine-tuning software
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
What Restrictions and Policies Are Attached to Inkling?
It is not yet fully clear how the separate Model Acceptable Use Policy (AUP) interacts with the open weights license. Reports suggest restrictions on surveillance, deception, and automated decision-making, but the exact scope and enforceability of these restrictions are not confirmed. The transparency of these policies and their practical impact on users remains uncertain.
From Weights to Wisdom: The Complete Guide to Running and Adapting Opensource AI Models
As an affiliate, we earn on qualifying purchases.
As an affiliate, we earn on qualifying purchases.
Next Steps for Testing and Adoption of Inkling
Independent researchers and organizations are expected to begin testing Inkling’s capabilities across various domains, including multimodal tasks and safety evaluation. Further benchmark results and real-world deployments will clarify its performance relative to other models. Additionally, scrutiny of the AUP and licensing terms will influence how broadly the model is adopted and integrated into commercial and research workflows. The company may also release updates or clarifications on usage restrictions.
As an affiliate, we earn on qualifying purchases.
Key Questions
What makes Inkling different from other large AI models?
Inkling is notable for being openly available with full weights under a permissive Apache 2.0 license, supporting multimodal inputs, and being trained on diverse data, which provides greater transparency and ownership potential for users.
Are there restrictions on how I can use Inkling?
Yes, reports suggest that Thinking Machines has a separate Acceptable Use Policy that may restrict certain applications, such as surveillance or deception, but the full scope and enforceability are not yet confirmed.
How does Inkling compare in performance to other models?
In benchmark tests, Inkling excels in safety-related tasks and speech processing but is mid-pack or behind on some language understanding benchmarks. Its smaller variant, Inkling-Small, matches or exceeds larger models on several metrics.
Will the training data and pipeline be released?
No, the training data and full pipeline have not been disclosed, which limits full transparency despite the open weights.
What are the implications of open weights for AI safety?
Open weights promote transparency and independent testing, but without full data disclosure and clear policies, safety and misuse concerns remain significant and require ongoing scrutiny.
Source: ThorstenMeyerAI.com
