cayu.context.thinking
Public Cayu API. Definitions are grouped by their public source module, and definitions from private implementation modules appear on their package’s page; adding a symbol does not move existing links. The import shown at each entry is a supported public namespace. Read the Python API overview and install the required extra.
ThinkingConfig
Section titled “ThinkingConfig”from cayu.context import ThinkingConfigProvider-neutral thinking/reasoning configuration.
Effort names are native values, not a portable scale. Supported vocabulary is
none, minimal, low, medium, high, xhigh, and max;
adapters reject known incompatible model/transport combinations before dispatch.
Unknown model compatibility remains backend-dependent. xlow is not a verified
native value in the maintained adapters and is not an alias for minimal.
Explicit effort maps unchanged to OpenAI reasoning.effort, compatible Chat
Completions reasoning_effort, or Anthropic adaptive thinking plus
output_config.effort. Bedrock Converse has no typed effort mapping and rejects
it. See cayu guide thinking for the matrix.
none requests the backend’s native non-reasoning effort; minimal still
permits reasoning. Both require enabled=True (the default), which enables
this configuration, not a promise that the model thinks. Neither is silently
mapped to enabled=False. The latter retains its historical best-effort
behavior: Anthropic disables thinking; OpenAI and generic Chat Completions no-op.
effort and max_tokens are mutually exclusive. A token budget selects
Anthropic legacy thinking (OpenAI and generic Chat Completions historically
ignore it). Disabling thinking cannot be combined with either control. With
neither control, enabled thinking uses existing provider defaults. Typed effort
wins over conflicting raw effort; unrelated provider options remain intact.
Bases: BaseModel.
Fields and attributes
Section titled “Fields and attributes”Defaults below are source expressions; Field(...) includes schema/validation settings and default_factory creates a fresh value. Inherited declarations are included. Constructor behavior follows the model or dataclass shown in the bases/decorators.
| Name | Type | Default |
|---|---|---|
enabled |
StrictBool |
True |
effort |
ThinkingEffort | None |
None |
max_tokens |
StrictInt | None |
None |
include_in_transcript |
StrictBool |
True |
validate_max_tokens (method)
Section titled “validate_max_tokens (method)”ThinkingConfig.validate_max_tokens(value: int | None) -> int | Nonevalidate_mode (method)
Section titled “validate_mode (method)”ThinkingConfig.validate_mode() -> ThinkingConfig