model.context_factory

M.vocab

Return the model's vocabulary as an ion7.core.Vocab handle. Cached on the Model — repeated calls are O(1).

M.context

Create an inference context from this model.

raises — When context creation fails after the retry cascade.

M.embedding_context

Create an embedding context : pooling enabled, no logits, CPU-only by default. Used for vector-search workloads where the model only emits a single vector per input rather than token-by-token output.

raises — When creation fails.