German Companies Rapidly Shift to Open-Weight AI Models Amidst Market Dynamics in 2026
German companies significantly increased their use of open-weight AI models in 2026, rapidly switching providers amid market shifts and emerging security concerns.
Key details
- Open-weight AI models dominated with 71.5% token volume at AI Gateway in September 2026.
- Revenue share of open-weight models increased to 23%, up from 5% in June 2026.
- Nearly 48% of teams switched their main AI model and 28.6% changed providers in September 2026.
- Emerging security risks from autonomous AI agents require continuous monitoring solutions like Microsoft's Perception.
In September 2026, open-weight AI models accounted for 71.5% of token volumes processed through the AI Gateway by Vercel, marking a dominant presence in the German AI market. The revenue share of these models rose significantly to 23%, jumping from just 5% in June 2026, reflecting an accelerated adoption trend among German enterprises. Spending on open-weight models surged by more than 70 times, contrasting with a tenfold increase for proprietary frontier models.
Market dynamics revealed a decline in the share of proprietary models such as Anthropics, whose gateway expenditure share dropped from 63.7% the previous month to 43.9%. Meanwhile, the DeepSeek-V4.1-Flash model processed nearly half of all open-weight tokens in September but generated only 2.6% of the spending, indicating lower token pricing. Additionally, the latest open-weight model, Jev from TypeSafe AI, quickly gained traction, capturing 35.3% of paying teams within two weeks, albeit accounting for a minor revenue share.
This shift is accompanied by high fluidity; in September, nearly 48% of teams switched their most used AI models, and 28.6% changed AI providers, underlining increased interchangeability and competitive pressure in the AI service market.
Despite rising popularity of Chinese AI models, German businesses face potential uncertainty due to recent export controls proposed on October 16, especially affecting US-based distribution of these models.
In parallel, companies deploying autonomous AI agents must address emerging insider security risks. Microsoft’s Perception platform offers continuous monitoring and dynamic risk evaluation of such agents, emphasizing the need for comprehensive security spanning identities, data, and networks to manage increasing operational and cyber risks.
Overall, German firms are dynamically adopting and optimizing open-weight AI technologies in a rapidly evolving landscape while balancing efficiency gains with emerging security challenges.