SPEC DATASHEET // llama-4-maverick
Meta AI/RELEASED Apr 18, 2026
Llama 4 Maverick (128E)
OPEN WEIGHTS SOTAMeta's flagship open-weights foundation model. 128-expert MoE architecture with 1M token context, powering enterprise on-premise deployments and community fine-tuning.
Context
1,048,576 tokens
Architecture
128-Expert Mixture-of-Experts
Pricing (1M)
Free / Open
License
Meta Community License
Verified Benchmark Suite
SWE-BENCH
71.2%
MMLU-PRO
83.6%
Quickstart Snippet
curl -s https://modelregistry.tirup.in/api/cli?model=llama-4-maverick