OVHcloud AI Endpoints

OVHcloud is the leading cloud provider in Europe and provides AI Endpoints: inference APIs for a selection of open-source models, such as Llama, Qwen, GPT OSS, and more. All inferences run in Europe, offering GDPR compliance, sovereignty, and data privacy. Your prompt and the LLM response are neither used nor saved.

Source: OVHcloud AI Endpoints

Model overview

OVHcloud AI Endpoints offers access to multiple open-weight models through a serverless inference API. The recommended model is gpt-oss-120b, a powerful 120 billion parameter model made by OpenAI.

You can also explore our catalog to browse all our available models.

Setup

Quick start

Configure via CLI:

clawdbot onboard --auth-choice ovhcloud-api-key

Or set the API key manually:

export OVHCLOUD_API_KEY="your-api-key-here"

Config snippet

{
  agents: {
    defaults: {
      model: { primary: "ovhcloud/gpt-oss-120b" }
    }
  }
}

Notes

Model refs use ovhcloud/<model> format.
Your data is not reused or kept by OVHcloud; data privacy is guaranteed.
Update pricing values in models.json if you need exact cost tracking.
See Model providers for provider rules.
Use clawdbot models list and clawdbot models set ovhcloud/gpt-oss-120b to switch models.

Troubleshooting

"Unknown model: ovhcloud/model-name"

This usually means the OVHcloud provider isn't configured (no provider entry and no OVHcloud auth profile/env key found). Fix by:

Running clawdbot configure and selecting OVHcloud, or
Adding the models.providers.ovhcloud block manually, or
Setting OVHCLOUD_API_KEY (or an OVHcloud auth profile) so the provider can be injected.

Make sure the model id is case-sensitive: ovhcloud/gpt-oss-120b.

Then recheck with:

clawdbot models list

2.1 KiB Raw Blame History