
No ratings yet
Be the first to review this model
Meta's natively multimodal open-weight model with an unprecedented 10 million token context window, fitting on a single NVIDIA H100 GPU. Features 16 experts with 17B active parameters for efficient inference at massive scale. More powerful than all previous Llama generations while remaining practical to deploy.
Released
April 5, 2025
Parameters
109B (MoE, 17B active)
Context
10M
Pricing
Open Source
Last updated: March 15, 2026
Benchmark scores may vary based on evaluation methodology and conditions.