SiliconData: Inference Token Spend Has Been Shifting Away From Closed Models Since Mid-July, Narrowing the Open/Closed Gap

TechRepublic (SiliconData data via Financial Times; corroborated by Silicon Data's own published index commentary)·medium signalXBlueskyLinkedInCopy link

TechRepublic reported 2026-08-14 that SiliconData figures, via the Financial Times, show a measurable shift in enterprise inference token spend away from closed models since mid-July, with the proprietary-versus-open gap narrowing over the past month even though absolute proprietary spend remains far higher. SiliconData's own analysis adds a counterintuitive wrinkle: effective prices for proprietary models fell sharply as OpenAI cut rates, while effective prices for open models actually rose as stronger Chinese near-frontier models like GLM-5.2 and Kimi K3 were served at premium rates. The convergence is happening from both directions, which undercuts the simple 'open models are the cheap option' framing.