Llama 4 Maverick
meta/llama-4-maverickThe open flagship of the Llama 4 generation, multimodal, million-token, permissively licensed.
1M ctx · Free
Meta
Open weights at the frontier (Llama)
Models
2
Founded
2004
Country
United States
Menlo Park
Founded 2004. FAIR/Llama programme from 2023.
Meta builds its models out of Menlo Park, United States. Webparam tracks two of its models, spanning language: free to call and reaching 10M. Every one of them ships with .
Models
The Meta catalogue
Ordered by weekly usage on Webparam.
Llama 4 Maverick by Meta MetaOn request Llama 4 Scout by Meta MetaOn request Llama 4 Scout
meta/llama-4-scoutTen million tokens of context, the largest window on the market, open weights, commodity prices.
10M ctx · Free
Recent releases
What shipped lately
History
Meta made open weights a frontier strategy. Llama downloads number in the hundreds of millions, and the Llama 4 generation pushed open models to 10M-token context, the largest window on the market. Much of the world's fine-tuned model ecosystem stands on Llama foundations.
Meta made open weights a frontier strategy. Llama downloads number in the hundreds of millions, and the Llama 4 generation pushed open models to 10M-token context, the largest window on the market. Much of the world's fine-tuned model ecosystem stands on Llama foundations.
Strengths
- The most-adopted open weights in the world
- 10M-token context (Scout)
- Massive fine-tuning ecosystem
Use Meta models through Webparam
Every model above is callable through one OpenAI-compatible convention: change the model ID, keep the request.
Chat completions reference →Llama 4 Maverick model details →
Related providers