Context window
1,000,000 tokens
Checked 19 Oct 2026https://github.com/meta-llama/llama-models/blob/main/models/llama4/MODEL_CARD.md
LLM · Meta
Llama 4 Maverick
Meta's Llama 4 model card lists Maverick as 17B active and 400B total parameters, with 128 experts and a 1M context length. Weights are released. The card does not publish a first-party token price. Knowledge cutoff on the card is August 2024. The llama-models README at https://github.com/meta-llama/llama-models/blob/main/README.md, checked 19 October 2026, still lists Llama 4, launched 5 April 2025, as the latest generation.
Checked 19 Oct 2026https://github.com/meta-llama/llama-models/blob/main/models/llama4/MODEL_CARD.md
Input price
No first-party API price
Checked 19 Oct 2026https://github.com/meta-llama/llama-models/blob/main/models/llama4/MODEL_CARD.md
Output price
No first-party API price
Checked 19 Oct 2026https://github.com/meta-llama/llama-models/blob/main/models/llama4/MODEL_CARD.md
API deployment
Not published
Checked 19 Oct 2026https://github.com/meta-llama/llama-models/blob/main/models/llama4/MODEL_CARD.md
Local / self-hosted
Yes
Checked 19 Oct 2026https://github.com/meta-llama/llama-models/blob/main/models/llama4/MODEL_CARD.md
VPC / cloud native
Not published
Checked 19 Oct 2026https://github.com/meta-llama/llama-models/blob/main/models/llama4/MODEL_CARD.md
License
Llama 4 Community License
Checked 19 Oct 2026https://github.com/meta-llama/llama-models/blob/main/models/llama4/LICENSE
Data retention
Not published
Checked 19 Oct 2026https://github.com/meta-llama/llama-models/blob/main/models/llama4/MODEL_CARD.md
Parameters
17B active / 400B total
Checked 19 Oct 2026https://github.com/meta-llama/llama-models/blob/main/models/llama4/MODEL_CARD.md