Overview
OLMo 2 is part of AI2’s effort to build fully open models, publishing not only weights but the data, training code and evaluation, so researchers can study every stage of training.
Highlights
- 7B and 13B sizes (larger versions followed)
- Training data and code released
- Apache 2.0 license
- Designed for research reproducibility
Typical uses
Research, education and applications that require full transparency about training data.
Things to consider
- Short 4K context in the original release.
- Download sizes refer to the default 4-bit quantized versions in the Ollama library; full-precision weights are larger.
Specifications are taken from the model card on Hugging Face, the Ollama library and the provider’s announcement. Benchmark results depend on evaluation settings; see how AI benchmarks work.