Run OpenCode agents with any inference provider
OpenCode is an open-source AI coding agent from SST that works with multiple inference providers. With SAM, you can run OpenCode on your cloud VMs using Scaleway, Google Vertex, Anthropic, or any OpenAI-compatible API as the backend.
Why use OpenCode with SAM
Provider Flexibility
Use any OpenAI-compatible inference provider — Scaleway, Google Vertex, Anthropic, or your own endpoint.
Open Source
OpenCode is fully open source (from SST), so you can audit, modify, and extend the agent to fit your needs.
Lightweight Footprint
Minimal resource requirements mean you can run more OpenCode instances on smaller VMs.
Scaleway Native
First-class integration with Scaleway's inference API — a natural pairing with SAM's Scaleway cloud provider support.
Get started in four steps
Configure Provider
Set your inference provider credentials — Scaleway API key, Google Vertex config, or a custom endpoint.
Select OpenCode
Choose OpenCode as the agent for your project.
Submit a Task
Describe your coding task. SAM provisions a VM and launches OpenCode.
Agent Delivers
OpenCode completes the work, and SAM commits and pushes the changes.
What you can build
Self-hosted AI stack
Pair OpenCode with a self-hosted inference endpoint for a fully private AI coding setup — no data leaves your infrastructure.
Scaleway-native development
Use Scaleway for both compute (VMs) and inference (Scaleway AI) — a unified European cloud stack.
Custom model experimentation
Test different models and providers by swapping OpenCode's inference backend without changing your workflow.
Frequently asked questions
What inference providers work with OpenCode?
OpenCode supports Scaleway, Google Vertex AI, Anthropic, and any OpenAI-compatible endpoint. Configure the provider in your project settings.
Is OpenCode the same as OpenAI Codex?
No — OpenCode is a separate open-source project from SST. It's a different agent that happens to support multiple AI providers including (but not limited to) OpenAI-compatible APIs.
Can I use my own fine-tuned models?
Yes — if your model is served via an OpenAI-compatible API endpoint, OpenCode can use it. Point the inference URL to your custom endpoint.
Start running OpenCode on your infrastructure
Self-host on Cloudflare's free tier. Bring your own cloud. Your agents, your code.