The model market in 2026: gaps shrink, licences decide
AI · not reviewedWritten with AI from the sources listed, not yet editorially reviewed.
According to the Stanford AI Index 2026, the leading US and Chinese models are only 2.7% apart — for businesses, licence, operation and data control now matter more than the last benchmark point.
The essentials in five points
- The performance gap between the leading US and Chinese models shrank to 2.7% — in May 2023 it was still 17.5 to 31.6 percentage points.
- The leading closed model is ahead of the leading open model by 3.3% — slightly more than in August 2024 (0.5%).
- In 2025, 50 notable models came from the US and 30 from China; China’s count doubled from the previous year.
- Licences differ widely: Qwen3 and Mistral Small 3.1 are under Apache 2.0, DeepSeek-V4 under MIT; Llama 4 and Gemma 3 use custom licences with conditions.
- According to the AI Index, 88% of organisations report using AI.
What this means for businesses
When the top models are close together, other criteria decide: where does the data run, which licence permits which use, what does operation cost, and how easily can you switch models? Build applications so the model stays swappable, and have licences legally reviewed before going into production.
Sources · 2
- The Next Web (Stanford AI Index 2026) · 19 Apr 2026Stanford AI Index 2026: China–US performance gap
- Hugging Face (Community-Blog) · 13 May 2026Open-Source- und Open-Weight-Modelle für den lokalen Betrieb 2026 (Übersicht)
Links open the original source in a new tab. Figures from company statements are labelled as vendor or company figures in the text.
Matching hardware
Private AI on premises
Run language models in-house so that confidential data never leaves your organisation.
More from this branch
All in Models →The next OROBO Report comes to you.
Planned from December 2026, every Thursday. Free, unsubscribe in one click.