LocalForge
Back to guide
MODEL / META

Llama 4 Scout

Meta's single-H100 fit requires INT4. All 109B parameters must be stored; 17B active is not the memory number.

Primary source
Total parameters109B17B active / token
Estimated memory~78 GBINT4 MoE
Context10M advertisedText + Image
LicenseLlama 4 CommunityMultilingual
01 / BEST FOR
  • long context
  • vision
  • datacenter serving
02 / ESTIMATED FIT
03 / DEPLOY WITH
vLLMAdvancedvllm serve meta-llama/Llama-4-Scout-17B-16E-Instruct --dtype auto
TensorRT-LLMAdvancedtrtllm-serve meta-llama/Llama-4-Scout-17B-16E-Instruct
llama.cppMediumllama-server -m /path/to/verified-model.gguf --port 8080
Watch out

Meta's single-H100 fit requires INT4. All 109B parameters must be stored; 17B active is not the memory number.