Scaleway Generative APIs, Managed Inference & GPU Instances

Offering overview

Developers who want European-hosted AI compute can pick from three tiers at Scaleway. Generative APIs serve open models such as Mistral Small 3.2, gpt-oss-120b, Llama 3.3 70B, Qwen3 variants, DeepSeek V4 Flash, Gemma 4 and GLM on a per-token basis from the Paris region. Managed Inference provisions a dedicated L4, L40S or H100 endpoint for predictable throughput, and GPU Instances expose L4, L40S, H100 PCIe, H100 SXM and B300 SXM accelerators for training and custom stacks. Useful extras include a free tier of one million tokens, a Batches API at a 50 percent discount, and the ability to keep workloads inside a single French zone (PAR-2) for compliance reasons. Hardware is provisioned hourly with no long-term commitment required. The lineup fits startups and enterprises that prefer EU jurisdiction and transparent euro pricing. Mistral Small 3.2 24B costs EUR 0.15 per 1M input tokens, embeddings start at EUR 0.10, and a dedicated L4 inference node is EUR 0.93 per hour.

Technical specifications

Containers

Unknown

Kubernetes

Unknown

Root access

Unknown

API access

Unknown

Autoscaling

Unknown

Persistent processes

Unknown

AI capabilities

Model serving

Mistral Small 3.2 24B, Mistral Medium 3.5, Pixtral 12B, gpt-oss-120b, Llama 3.3 70B, Qwen3 235B, Qwen3 Coder 30B, Qwen3.5 397B, Qwen3.6, Qwen3.8, DeepSeek V4 Flash, Gemma 4 26B, GLM 5.2, Qwen3 Embedding 8B, bge-multilingual-gemma2, Whisper large-v3

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 AI hosting options does Scaleway provide?

Scaleway offers serverless Generative APIs for open models, Managed Inference with dedicated GPU endpoints, and GPU Instances for training and custom software stacks.

Which GPUs support Scaleway's AI services?

Managed Inference offers L4, L40S and H100 endpoints. GPU Instances offer L4, L40S, H100 PCIe, H100 SXM and B300 SXM accelerators.

Where does Scaleway host its Generative APIs?

Generative APIs serve models from the Paris region. Scaleway also offers the ability to keep workloads inside the French PAR-2 zone.

Does Scaleway require a long-term hardware commitment?

Hardware is provisioned hourly, with no long-term commitment required.