Thinking Machines Lab Drops Its First Model

**Thinking Machines Lab** has launched its initial model, **Inkling**, as reported by **Wired**. The release centers on a **975-billion-parameter** **open source model** built for multimodal processing.

From a technical standpoint, **Inkling** was trained directly to interpret both **video and audio** inputs. Operating with **975 billion parameters**, the **open source model** focuses its product mechanics on large-scale multimodal comprehension across combined visual and auditory data streams.

For engineers and builders, the availability of a **975-billion-parameter** **open source model** offers direct access to inspect and deploy advanced **video and audio** capabilities. This provides technical teams with a large-scale multimodal foundation that can be run and adapted without relying solely on managed services.

In terms of market context, this first release allows **Thinking Machines** to establish a direct presence alongside major industry competitors such as **Anthropic** and **OpenAI**. Releasing **Inkling** openly gives **Thinking Machines Lab** a distinct positioning strategy in a landscape populated by closed proprietary models.

Practically, technical teams should evaluate how **Inkling** handles specific **video and audio** ingestion workflows. The primary dynamic to watch next is whether **Thinking Machines** can build sustained adoption around its **open source model** relative to established options from **OpenAI** and **Anthropic**.