The scaleway provider targets the OpenAI-compatible API exposed by Scaleway AI. It works seamlessly with AI Kit agents and workflows.
Installation
Configuration
Retrieve the key from the Scaleway console.
Use with an agent
Available models
Examples
Structured output
Scaleway still validates schemas on the client pipeline. Type your schema with AgentStructuredOutput to keep inference and DX identical to OpenAI/Google.
Inside a workflow
Streaming
Reasoning effort
Scaleway reasoning models expose the OpenAI-compatible reasoning_effort parameter. With AI Kit, pass it through providerOptions.scaleway.reasoningEffort.
Scaleway accepts these values: "none", "low", "medium", "high".
Use reasoningEffort, not reasoningLevel. The AI SDK maps reasoningEffort to the HTTP reasoning_effort field for OpenAI-compatible providers.
Model selection tips
- Code generation:
qwen3-coder-30b-a3b-instruct, devstral-small-2505.
- General-purpose tasks:
gpt-oss-120b, llama-3.3-70b-instruct.
- Lightweight workloads:
llama-3.1-8b-instruct, mistral-nemo-instruct-2407.
- Reasoning-heavy tasks:
qwen3.5-397b-a17b, deepseek-r1-distill-llama-70b, qwen3-235b-a22b-instruct-2507.
- Multimodal scenarios:
pixtral-12b-2409.
Best practices
- Security – store API keys in a secret manager, never commit them.
- Error handling – wrap calls in
try/catch and log failures.
- Cost control – set
maxOutputTokens and monitor usage.
- Temperature – pick the right creativity level (
0.0-0.3 precise, 0.4-0.7 balanced, 0.8+ creative).
Resources