OVHcloud has launched AI Endpoints, a serverless solution designed to make open-source artificial intelligence models more accessible to developers and businesses. The platform offers over 40 models, including large language models (LLMs) and generative AI tools, supporting applications such as chatbots, speech transcription, and code generation.
With AI Endpoints, developers can integrate advanced AI capabilities into their applications without needing to manage infrastructure or possess deep machine learning expertise. The service is hosted on OVHcloud’s trusted cloud environment, allowing users to experiment with and deploy AI models securely and efficiently.
Support for diverse business use cases
AI Endpoints provides a sandbox environment for developers to test features before rolling them out across applications and business processes. The platform is suited for a wide range of AI applications, including real-time conversational agents, data extraction, speech recognition and synthesis, and coding assistance.
For example, LLMs can be embedded into applications to enhance customer service or user interaction. Text extraction capabilities help businesses process unstructured data, improving operational workflows. Through voice APIs, developers can incorporate both transcription and voice response features, supporting voice-based user interfaces. Additionally, coding tools such as Continue offer in-IDE support with code suggestions and error detection to streamline development.
Privacy, transparency, and environmental responsibility
The platform is built on OVHcloud’s energy-efficient infrastructure, which relies on water-cooled servers housed in environmentally friendly data centres. This approach helps minimise the environmental footprint of AI operations while maintaining performance.
A key differentiator of AI Endpoints is its focus on data sovereignty and transparency. By hosting the solution in Europe, OVHcloud ensures data is protected from non-European regulations. The use of open-weight AI models allows organisations to migrate or replicate these models across different infrastructures, giving them greater control over their data and applications.
“We are excited to launch AI Endpoints and are humbled by the incredible feedback we get from our amazing community. With support for the most diverse and sought after open source LLM models, AI Endpoints helps to democratise AI so developers can add to their apps the most cutting-edge models. Our solution enables them to do this easily in a trusted cloud environment with full confidence in OVHcloud’s sovereign infrastructure,” said Yaniv Fdida, Chief Product and Technology Officer at OVHcloud.
Flexible pricing and regional availability
Following an early preview phase, AI Endpoints is now live in Asia-Pacific, Canada, and Europe, with services deployed from OVHcloud’s Gravelines data centre. Based on user feedback, the service includes enhanced features such as better API key management, increased model stability, and a wider range of supported models.
The offering covers several model categories, including LLMs like Llama 3.3 70B and Mixtral 8x7B; small language models such as Mistral Nemo and Llama 3.1 8B; code models like Qwen 2.5 Coder 32B and Codestral Mamba; reasoning tools such as DeepSeek-R1; multimodal models like Qwen 2.5 VL 72B; image generation with SDXL; and speech-to-text (ASR) and text-to-speech (TTS) capabilities.
Pricing is offered on a pay-as-you-go basis, with costs calculated by the number of tokens processed per minute, depending on the selected model.
