Atlas · frontier
Open weights
Some labs publish the model file itself, so anyone can run it. A weights release does not by itself include the data or code.
In plain words
An open-weights model is one whose numbers you can download and run yourself. Other models are offered only as a service, with no weights to download.
How it works
Meta releases Llama 4 Scout in BF16, and Maverick in BF16 and FP8. Mistral calls Large 4 an open-weight model, under its own Mistral Ecosystem Open Weight License. A release of weights does not by itself say which training data or code are available. A report like Meta’s Llama 3 paper, which states 405B parameters trained on 15.6T tokens, is how you learn what went in.
Going deeper
Ollama can run models imported in its supported formats, such as GGUF. The Frontier shows, dated, which current models publish weights and sizes: of the eight listed, only Meta’s and Mistral’s do.
Sources
- Llama 4 model card Meta
Llama 4 Scout, a 17 billion parameter model with 16 experts, and Llama 4 Maverick, a 17 billion parameter model with 128 experts.
400B (Total)
109B (Total)
Llama 4 Maverick (17Bx128E)
the Llama 4 Maverick model is released as both BF16 and FP8 quantized weights.
The Llama 4 Scout model is released as BF16 weights
The pretraining data has a cutoff of August 2024.
Model Release Date: April 5, 2025
Since Meta is openly releasing these models
- Mistral Large 4 Mistral
Mistral Large 4 is a state-of-the-art, open-weight, general-purpose multimodal model with a granular Mixture-of-Experts architecture.
It features 52B active parameters and 1.05T total parameters, and a 1.6B vision encoder.
Mistral Ecosystem Open Weight License
v26.10
- The Llama 3 Herd of Models arXiv
Specifically, we pre-trained a flagship model with 405B trainable parameters on 15.6T text tokens.
We pre-train Llama 3 on a corpus of about 15T multilingual tokens, compared to 1.8T tokens for Llama 2.
- Importing a Model Ollama
Ollama does not quantize GGUF models during import. Prepare and quantize them first with a GGUF tool such as llama.cpp’s llama-quantize.
Quotes were checked against the live pages on 2026-11-11.