Just two weeks after launching its first open-source language model, Inkling, startup Thinking Machines has already followed up with a leaner sibling. According to VentureBeat, the new Inkling-Small model retains nearly all of its predecessor's capabilities while running at approximately one-quarter of the size, and in some benchmarks it even edges past the original.

The company, founded by former OpenAI chief technology officer Mira Murati, has built a reputation for moving quickly since emerging from stealth. Releasing a smaller, more efficient model so soon after the first suggests a deliberate strategy: proving that strong performance doesn't have to come bundled with heavy compute costs. For businesses evaluating open-source AI, that combination of speed and efficiency is increasingly the deciding factor.

Smaller models that preserve most of their larger counterparts' abilities matter well beyond research labs. They lower the barrier for companies wanting to run capable language models on more modest infrastructure, whether on-premises, at the edge, or in cost-sensitive cloud deployments. As open-source AI continues to mature at this pace, organizations gain more practical, affordable options for building their own AI-powered products.

At Generative AI Solutions, we see this kind of rapid, efficient iteration as a model for how quickly great ideas can become working products.