Scaleway on Hugging Face Inference Providers 🔥

Source: HuggingFace Date: 2025-09-19 URL: https://huggingface.co/blog/inference-providers-scaleway

Summary

Integration announcement: Scaleway (European cloud) joins HF Inference Providers. €0.20/million tokens, sub-200ms first token latency, European data centers (Paris) for data sovereignty. Supports structured outputs, function calling, and multimodal inference. Available models include gpt-oss-120b, Qwen3-Coder-30B, DeepSeek R1 Distill Llama-70B, Gemma 3 27B. Accessible via HF UI, huggingface_hub Python SDK, and @huggingface/inference JS SDK.

Implications

HF as open-source ML hub. Scaleway as a European inference provider addresses a data sovereignty requirement that US-based providers cannot fulfill for EU-regulated industries. The HF Inference Providers model — route through HF with no markup, billed to HF account — means European organizations get a GDPR-compliant path to open-weights inference through the same SDK they already use.

Open-weights ecosystem health. €0.20/million tokens is competitive with Together AI and Groq pricing for comparable models. Adding EU-based capacity to the inference provider ecosystem reduces the geographic concentration of open-weights inference infrastructure — a resilience improvement that matters as regulatory requirements for data residency tighten.