ooai_llm.profiles

Serializable chat-model profiles and runtime wrapper.

Purpose:

Provide a durable Pydantic profile that can be checked into configuration and used to create LangChain chat models through the existing factory surface. The LLM runtime owns one profile, lazily builds the runnable, and records observed usage/cost metadata after calls.

Attributes

CacheBackendName
logger

Classes

CacheKeyPolicy	Serializable cache policy for a chat-model profile.
ChatModelProfileResolution	Resolved model/profile metadata without constructing a chat model.
ChatModelProfile	Serializable configuration for creating a LangChain chat model.
LLM	Stateful runtime wrapper around a serializable ChatModelProfile.

Module Contents

ooai_llm.profiles.CacheBackendName

ooai_llm.profiles.logger

class ooai_llm.profiles.CacheKeyPolicy(/, **data: Any)

Bases: pydantic.BaseModel

Serializable cache policy for a chat-model profile.

model_config

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

enabled: bool = True

namespace: str = 'default'

key: str | None = None

backend: CacheBackendName | None = None

class ooai_llm.profiles.ChatModelProfileResolution(/, **data: Any)

Bases: pydantic.BaseModel

Resolved model/profile metadata without constructing a chat model.

model_config

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

model: ooai_llm.types.ModelString

profile_id: str | None = None

metadata: ooai_llm.metadata.ModelInfo

reasoning: ooai_llm.reasoning.ReasoningResolution | None = None

cache_namespace: str | None = None

cache_key: str | None = None

class ooai_llm.profiles.ChatModelProfile(/, **data: Any)

Bases: pydantic.BaseModel

Serializable configuration for creating a LangChain chat model.

model_config

Configuration for the model, should be a dictionary conforming to [ConfigDict][pydantic.config.ConfigDict].

id: str | None = None

model: str | None = None

alias: ooai_llm.settings.ModelAliasName | None = None

provider: str | ooai_llm.providers.Provider | None = None

preset: ooai_llm.settings.ModelPresetName = 'default'

temperature: float | None = None

max_tokens: int | None = None

top_p: float | None = None

frequency_penalty: float | None = None

presence_penalty: float | None = None

seed: int | None = None

stop: str | list[str] | None = None

max_retries: int | None = None

parallel_tool_calls: bool | None = None

timeout: float | None = None

streaming: bool | None = None

configurable_fields: str | list[str] | None = None

config_prefix: str | None = None

model_kwargs: dict[str, Any] = None

constructor_kwargs: dict[str, Any] = None

reasoning: ooai_llm.reasoning.ReasoningInput = None

auto_refresh_models: bool | None = None

force_model_refresh: bool = False

billing_model_name: str | None = None

cache: CacheKeyPolicy | bool | None = None

tags: list[str] = None

metadata: dict[str, Any] = None

run_name: str | None = None

cost_labels: dict[str, str] = None

to_json(*, indent: int = 2) → str

Serialize the profile to deterministic JSON.

classmethod from_json(text: str) → ChatModelProfile

Deserialize a profile from JSON text.

create_llm(*, settings: ooai_llm.settings.AppSettings | None = None, **kwargs: Any) → Any

Create a LangChain chat model from this profile.

create_bundle(*, settings: ooai_llm.settings.AppSettings | None = None, messages: Any = None, tools: collections.abc.Sequence[Any] | None = None, **kwargs: Any) → ooai_llm.metadata.CreatedLLMBundle

Create a chat model plus resolved metadata from this profile.

create_runtime(*, settings: ooai_llm.settings.AppSettings | None = None, recorder: ooai_llm.callbacks.UsageRecorder | None = None, budget: ooai_llm.callbacks.BudgetPolicy | None = None, id: str | None = None, uuid: str | uuid.UUID | None = None) → LLM

Create an LLM runtime from this profile.

create_dspy_lm(*, settings: ooai_llm.settings.AppSettings | None = None, **kwargs: Any) → Any

Create a DSPy LM from this profile.

DSPy is an optional dependency and is imported lazily. Install ooai-llm[dspy] before using this helper.

resolve(*, settings: ooai_llm.settings.AppSettings | None = None, auto_refresh_models: bool | None = None, force_model_refresh: bool | None = None) → ChatModelProfileResolution

Resolve model, metadata, reasoning, and cache key material.

cache_key() → str

Return the explicit or stable derived cache key for this profile.

class ooai_llm.profiles.LLM(*, profile: ChatModelProfile | collections.abc.Mapping[str, Any], settings: ooai_llm.settings.AppSettings | None = None, recorder: ooai_llm.callbacks.UsageRecorder | None = None, budget: ooai_llm.callbacks.BudgetPolicy | None = None, id: str | None = None, uuid: str | uuid.UUID | None = None)

Stateful runtime wrapper around a serializable ChatModelProfile.

profile

settings

recorder

budget = None

uuid

id

property bundle: ooai_llm.metadata.CreatedLLMBundle

Return the lazily built bundle.

property runnable: Any

Return the lazily built LangChain runnable.

property model: ooai_llm.types.ModelString

Return the resolved model string.

property metadata: ooai_llm.metadata.ModelInfo

Return resolved model metadata.

property usage_events: list[ooai_llm.callbacks.UsageEvent]

Return recorded usage events.

property total_tokens: int

Return total observed tokens.

property total_cost_usd: decimal.Decimal

Return total observed/estimated cost.

property usage_summary: Any

Return grouped observed usage totals.

set_profile(profile: ChatModelProfile | collections.abc.Mapping[str, Any], *, id: str | None = None) → None

Replace the profile and invalidate the cached runnable.

invalidate() → None

Drop the cached runnable and bundle.

refresh(*, force: bool = False) → ooai_llm.metadata.CreatedLLMBundle

Rebuild the runnable, optionally forcing model-default refresh.

create_dspy_lm(**kwargs: Any) → Any

Create a DSPy LM from this runtime’s profile and settings.

The returned object is a native DSPy LM. The richer program/runnable layer belongs in ooai-agents.

invoke(input: Any, config: collections.abc.Mapping[str, Any] | None = None, **kwargs: Any) → Any

Invoke the underlying chat model and record observed usage.

async ainvoke(input: Any, config: collections.abc.Mapping[str, Any] | None = None, **kwargs: Any) → Any

Async invoke the underlying chat model and record observed usage.

batch(inputs: collections.abc.Sequence[Any], config: collections.abc.Mapping[str, Any] | None = None, **kwargs: Any) → list[Any]

Batch invoke the underlying chat model and record observed usage.

async abatch(inputs: collections.abc.Sequence[Any], config: collections.abc.Mapping[str, Any] | None = None, **kwargs: Any) → list[Any]

Async batch invoke the underlying chat model and record observed usage.

stream(input: Any, config: collections.abc.Mapping[str, Any] | None = None, **kwargs: Any)

Stream from the underlying chat model.

async astream(input: Any, config: collections.abc.Mapping[str, Any] | None = None, **kwargs: Any)

Async stream from the underlying chat model.