OVHcloud AI Endpoints & Public Cloud GPU Instances

Offering overview

Two complementary products make up the AI hosting line at this French cloud group. AI Endpoints exposes more than twenty open-weight models (LLMs, embeddings, speech-to-text, vision) through an OpenAI-compatible API, while Public Cloud GPU Instances give teams raw accelerators such as the Quadro RTX 5000, A10, L4, L40S, A100 and H100 for training or self-managed serving. The endpoints service has a Base mode billed per input and output token and a Batch mode at half that rate, with a free playground, version-controlled model lifecycle, zero data retention, and ISO 27001, SOC 2 Type II and HDS health-data certification. Integrations cover Hugging Face, LiteLLM, Continue and similar tooling, and everything runs on infrastructure inside the European Union. It suits organisations that want sovereignty guarantees without hyperscaler lock-in. Pricing is transparent: the gpt-oss-20b endpoint starts at EUR 0.04 per 1M input tokens, and the cheapest GPU instance (rtx5000-28) is listed at roughly USD 0.60 per hour.

Technical specifications

Containers

Unknown

Kubernetes

Unknown

Root access

Unknown

API access

Unknown

Autoscaling

Unknown

Persistent processes

Unknown

AI capabilities

Model serving

Llama 3.3 70B, Qwen3.5, Qwen3.6, Qwen3.8, Qwen2.5-VL 72B, gpt-oss-120b, gpt-oss-20b, Qwen3 Embedding 8B, bge-m3, Whisper large-v3, Stable Diffusion XL

AI frameworks

Not stated

Inference endpoints

Unknown

Fine-tuning

Unknown

Vector database

Unknown

Model catalog

Unknown

Monitoring

Not stated

Evaluation

Not stated

Security & compliance

Geography

Provider headquarters are separate from offering infrastructure and availability.

Available plans

Pricing can change. Promotional prices are shown separately from standard prices.

Questions & answers

What does OVHcloud AI Endpoints provide?

AI Endpoints exposes more than twenty open-weight models through an OpenAI-compatible API, covering language models, embeddings, speech-to-text and vision.

How do AI Endpoints differ from Public Cloud GPU Instances?

AI Endpoints provides model access through an API. Public Cloud GPU Instances provide accelerators such as the Quadro RTX 5000, A10, L4, L40S, A100 and H100 for training or self-managed serving.

Which tools integrate with OVHcloud AI Endpoints?

Documented integrations include Hugging Face, LiteLLM and Continue.

Where are OVHcloud's AI services hosted?

AI Endpoints and Public Cloud GPU Instances run on infrastructure inside the European Union.

What data retention policy is stated for AI Endpoints?

The AI Endpoints service is described as having zero data retention.