Thinking Machines bets on efficiency over size with its second model, Inkling Small
Thinking Machines released Inkling Small, an open-weights reasoning model. It is less than a third the size of its predecessor but outperforms it on coding and reasoning benchmarks.
Thinking Machines, founded by former OpenAI CTO Mira Murati, has launched its second large language model, Inkling Small. This release marks a strategic shift toward optimizing model efficiency rather than simply scaling parameter counts.
The new model is available with open weights and is designed as a reasoning system. According to the company, Inkling Small occupies less than one-third of the storage space of the original Inkling model while achieving superior results on specific coding and reasoning benchmarks.
This development highlights a growing trend in the AI sector where developers prioritize performance-per-parameter ratios. By demonstrating that smaller models can compete with or exceed larger predecessors, Thinking Machines aims to reduce inference costs and broaden accessibility for developers.
The lab has positioned itself as a competitor in the frontier model space since its inception. This second release suggests a maturing roadmap focused on practical utility and computational efficiency alongside raw capability.
This page provides an editorial summary based on publicly available information. It is not a republished article. Use the source link below for the original report.