The Pulse
Ecosia Says Its Mistral Exit Is About Open Models, Not China
Ecosia founder Christian Kroll says the search company is shifting from Mistral toward open AI models accessed through Germany’s Melious platform. He says the priority is cost, flexibility and energy use—not where a model was developed.
AI.info Team · Written with AI ·
“We’ll probably be using Melious.ai in future. Unlike Mistral, this is a European company that offers high-quality, affordable AI inference powered by green electricity,”
Christian Kroll, founder and CEO of Ecosia
Ecosia founder and CEO Christian Kroll says the search company’s move away from Mistral is about access to open models and the infrastructure running them, not a preference for Chinese technology. In an interview published by EU-Startups on October 9, Kroll said Ecosia would probably use Melious.ai, a European platform for AI inference. His wording points to a planned shift, not a claim that every Ecosia AI feature has already migrated.
Melious would serve models from several countries
The models named in the account include Alibaba’s Qwen, Z.ai’s GLM and Moonshot AI’s Kimi. All are developed in China, but Kroll’s stated focus is on the models being open and on the service infrastructure being European. Ecosia’s position, as described in the interview, is that model origin and the location where it runs are separate questions.
Melious says it runs open-weight models on European infrastructure powered by clean energy. The company’s CEO and co-founder, Simon Jakobs, told EU-Startups that customer data is not retained or used for training, and that the models cannot send data back to their developers when run on Melious’s servers. Those are Melious’s claims about its service, rather than independently measured figures in the interview.
Ecosia says it can switch models on sensitive subjects
Kroll also addressed concerns about political censorship in Chinese models. He said Ecosia can use alternative models for sensitive subjects, giving questions about Taiwan as an example, and that the company is working to reduce political bias in AI-generated answers. The account does not identify which model would handle any particular topic or explain how Ecosia tests for bias.
The approach makes model choice a service-level decision: Ecosia can select among available models while relying on Melious to provide the computing infrastructure. That may give the search company more room to change providers or models than a single-model arrangement, though the interview offers no details about the selection system or how often Ecosia expects to switch.
European hosting does not make every model European
Kroll’s clarification reframes the decision amid attention to the Chinese origin of some candidate models. Ecosia is not saying those models were developed in Europe; it is arguing that European hosting and data processing matter to its control over the service. The distinction leaves two separate questions: where a model was built, and who operates the infrastructure processing users’ queries.
Kroll also questioned Mistral’s European independence because of its reliance on foreign investors, particularly those from the United States. The interview sets out that concern alongside Ecosia’s preference for more affordable, flexible and energy-efficient open models, but gives no comparative cost or energy figures. For now, Kroll’s concrete commitment is limited: Ecosia will probably use Melious.ai in future, with open models running on European infrastructure.