The current balance of power in open models - Interconnects AI

Interconnects AI has published an analysis titled 'The current balance of power in open models.' The piece addresses open-weight artificial intelligence models. According to the analysis, open-weight models such as Meta's Llama, Alibaba's Qwen, Google's Gemma, and DeepSeek's models are the most common form of open models. The excerpt identifies those four model families by name as the most common examples of the category.

The analysis also states that leading Chinese models GLM-5.2 and Kimi K3 have enacted a step change in the commercial viability of open models. That claim is the central development described in the available excerpt, and it is attributed specifically to those two models. The excerpt does not provide figures, benchmarks, licensing terms, pricing, adoption data, or parameter counts alongside the claim. It also does not state which organisations produced GLM-5.2 or Kimi K3, when either was released, or what the step change consists of in measurable terms.

The excerpt frames the subject as a question of balance among open models, citing the four families as the most common form and the two Chinese models as the ones that shifted commercial viability. Interconnects AI published the analysis at interconnects.ai. Beyond the identification of the named models and the characterisation of the step change, the excerpt does not specify how the balance is measured, whether the assessment covers capability, cost, adoption, or licensing, or what evidence supports it. Those details would appear in the full analysis.