Enterprises looking to move more of their AI workloads to open-source models have a robust new option. Thinking Machines, an American AI startup founded by former OpenAI CTO Mira Murati, has unveiled Inkling. This open-source multimodal language model is designed to offer enterprises a customizable, censorship-resistant solution, and it is now available under the Apache 2.0 license. While it doesn’t claim the top spot in performance, Inkling brings a solid alternative to companies wary of the black-box nature of many AI models.
### What Inkling Does
Inkling is a multimodal Mixture-of-Experts (MoE) language model sporting 975 billion total parameters. It can process and reason across text, images, and audio, making it versatile for various applications. The model includes a “controllable thinking effort” mechanism, balancing cost and performance effectively. This makes it particularly appealing for businesses seeking a practical and adaptable AI solution without the overhead of more complex models.
For enterprises concerned about censorship, Inkling’s ability to handle controversial or sensitive topics without bias is a key selling point. Thinking Machines claims that the model delivers factual outputs regardless of controversy, which could attract organizations that prioritize transparency and control over their AI systems.
### Competitive Context
In the crowded landscape of 2026 open-weight AI models, Inkling faces stiff competition. Models such as GLM 5.2, DeepSeek V4 Pro, and Kimi K2.6 outperform Inkling in specific areas like coding, reasoning, and technical benchmarks. For instance, GLM 5.2 scores higher on coding and complex reasoning tasks, while DeepSeek excels in factuality domains. Kimi K2.6 outperforms Inkling in several technical benchmarks.
However, Inkling holds its ground in areas like mathematical problem-solving, where it surpasses DeepSeek V4 Pro. Its generalist approach offers a balanced performance across various benchmarks, appealing to enterprises that value versatility over specialization. While it doesn’t lead in every category, Inkling provides a competitive alternative for businesses that need a broad-spectrum AI model.
### Implications for Founders and Engineers
For founders and engineers, the release of Inkling represents an opportunity to explore open-source AI solutions that prioritize transparency and control. With the model’s open weights available on Hugging Face and an API through Thinking Machines’ Tinker platform, developers can integrate and experiment with the model in diverse environments.
The open-source nature of Inkling allows for customization and adaptation to specific enterprise needs, reducing dependency on proprietary models that may not align with an organization’s values or operational requirements. For engineers, this means more freedom to innovate and tailor AI solutions without being locked into a vendor’s ecosystem.
### What’s Next
Thinking Machines plans to continue developing and refining Inkling, with a preview of Inkling-Small already announced. This lighter version is optimized for low-latency and cost-sensitive applications, broadening the model’s appeal. As the AI landscape evolves, companies like Thinking Machines will need to focus on refining performance and expanding capabilities to stay competitive.
For founders and engineers, staying informed about these developments is crucial. The ongoing evolution of models like Inkling could impact the strategic decisions surrounding AI adoption and integration in enterprise environments. Keeping an eye on these advancements will help businesses leverage AI technologies effectively, ensuring they remain competitive in an increasingly AI-driven world.