Llama 4 Maverick

Llama 4 Maverick was released April 5, 2025 as a 17B active parameter mixture-of-experts model (400B total, 128 experts). Scored 92.2% on MMLU, outperforming GPT-4o and Gemini 2.0 Flash. Featured a 1M token context window and native multimodal support. Could be deployed on a single H100 DGX host.