An open-weight reasoning model designed for careful, inspectable local evaluations.
Q5_K_M default
tokens
Canonical metadata is separated from immutable version and Run evidence.
Aurora Reason 8B is an illustrative reasoning model with standardized and independently reproduced Runs across consumer GPUs and Apple Silicon.
2 published Runs; each result links to its exact conditions and evidence level.
Benchmark claims remain tied to immutable revisions even after the repository changes.
New reasoning curriculum and lower repetition under long contexts.
Explicit model relationships are stored as typed edges rather than inferred from names.
Other models used for reasoning.
A general-purpose multilingual instruction model with broad runtime support.
A tool-using agent model with strong planning but higher memory requirements.
