Parameters
109B MoE · 17B active · visionOpen Model Catalogue
Meta
Text generationVision
Llama 4 Scout
meta-llama/Llama-4-Scout-17B-16E-InstructMeta MoE with 10M context — 2 × H200
Min VRAM
≈ 250 GBHardware
4 × 80 GB GPUThis model requires accepting its licence on HuggingFace and a HuggingFace token in your account credentials.
OpenAI-compatible API with a per-workload key — drop-in for existing SDKs.
Dedicated GPUs — your model, your instance, no shared queues.
Billed by the second with auto-shutdown — never pay for idle time you didn't plan.