Skip to content
ZHIVEXDocs
DocumentationRelease notesGitHub

zhivex_ai.experimental

Python 0.23.0 · published wheel reference and guides.

PythonMixed

Public import paths and stability come from the installed artifact.

Use the root for Stable APIs and focused namespaces for extensions. Legacy root extension imports remain classified below.

AgentLiveEvent

Experimental · agent

from zhivex_ai.experimental import AgentLiveEvent
AgentLiveEvent (public type alias or constant; no callable signature)

LiveAgentStreamResult

Experimental · agent

from zhivex_ai.experimental import LiveAgentStreamResult
LiveAgentStreamResult(runner: 'asyncio.Task[AgentRunResult[AgentOutputT]]', broadcast: 'Broadcast[AgentLiveEvent]', live_session: 'asyncio.Future[Any]') -> 'None'
Abstract base class for generic types.

On Python 3.12 and newer, generic classes implicitly inherit from
Generic when they declare a parameter list after the class's name::

    class Mapping[KT, VT]:
        def __getitem__(self, key: KT) -> VT:
            ...
        # Etc.

On older versions of Python, however, generic classes have to
explicitly inherit from Generic.

After a class has been declared to be generic, it can then be used as
follows::

    def lookup_name[KT, VT](mapping: Mapping[KT, VT], key: KT, default: VT) -> VT:
        try:
            return mapping[key]
        except KeyError:
            return default

RealtimeAudioOutputEvent

Experimental · types

from zhivex_ai.experimental import RealtimeAudioOutputEvent
RealtimeAudioOutputEvent(type: "Literal['realtime-audio-output']" = 'realtime-audio-output', audio: 'bytes' = b'', media_type: 'str' = 'audio/pcm', sample_rate_hz: 'int | None' = None, channels: 'int | None' = None, item_id: 'str | None' = None, response_id: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>) -> None
RealtimeAudioOutputEvent(type: "Literal['realtime-audio-output']" = 'realtime-audio-output', audio: 'bytes' = b'', media_type: 'str' = 'audio/pcm', sample_rate_hz: 'int | None' = None, channels: 'int | None' = None, item_id: 'str | None' = None, response_id: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>)

RealtimeConnectOptions

Experimental · types

from zhivex_ai.experimental import RealtimeConnectOptions
RealtimeConnectOptions(timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None, metadata: 'dict[str, Any]' = <factory>, browser_client: 'bool' = False, subprotocols: 'list[str]' = <factory>) -> None
RealtimeConnectOptions(timeout_ms: 'int | None' = None, max_retries: 'int | None' = None, retry_backoff_ms: 'int | None' = None, metadata: 'dict[str, Any]' = <factory>, browser_client: 'bool' = False, subprotocols: 'list[str]' = <factory>)

RealtimeErrorEvent

Experimental · types

from zhivex_ai.experimental import RealtimeErrorEvent
RealtimeErrorEvent(type: "Literal['realtime-error']" = 'realtime-error', error: 'Exception | None' = None, message: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>) -> None
RealtimeErrorEvent(type: "Literal['realtime-error']" = 'realtime-error', error: 'Exception | None' = None, message: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>)

RealtimeEvent

Experimental · types

from zhivex_ai.experimental import RealtimeEvent
RealtimeEvent (public type alias or constant; no callable signature)

RealtimeModel

Experimental · types

from zhivex_ai.experimental import RealtimeModel
RealtimeModel(*args, **kwargs)
Base class for protocol classes.

Protocol classes are defined as::

    class Proto(Protocol):
        def meth(self) -> int:
            ...

Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).

For example::

    class C:
        def meth(self) -> int:
            return 0

    def func(x: Proto) -> int:
        return x.meth()

    func(C())  # Passes static type check

See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::

    class GenProto[T](Protocol):
        def meth(self) -> T:
            ...

RealtimeResponseCompletedEvent

Experimental · types

from zhivex_ai.experimental import RealtimeResponseCompletedEvent
RealtimeResponseCompletedEvent(type: "Literal['realtime-response-complete']" = 'realtime-response-complete', reason: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>) -> None
RealtimeResponseCompletedEvent(type: "Literal['realtime-response-complete']" = 'realtime-response-complete', reason: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>)

RealtimeSession

Experimental · types

from zhivex_ai.experimental import RealtimeSession
RealtimeSession(*args, **kwargs)
Base class for protocol classes.

Protocol classes are defined as::

    class Proto(Protocol):
        def meth(self) -> int:
            ...

Such classes are primarily used with static type checkers that recognize
structural subtyping (static duck-typing).

For example::

    class C:
        def meth(self) -> int:
            return 0

    def func(x: Proto) -> int:
        return x.meth()

    func(C())  # Passes static type check

See PEP 544 for details. Protocol classes decorated with
@typing.runtime_checkable act as simple-minded runtime protocols that check
only the presence of given attributes, ignoring their type signatures.
Protocol classes can be generic, they are defined as::

    class GenProto[T](Protocol):
        def meth(self) -> T:
            ...

RealtimeSessionConfig

Experimental · types

from zhivex_ai.experimental import RealtimeSessionConfig
RealtimeSessionConfig(instructions: 'str | None' = None, voice: 'str | None' = None, tools: "dict[str, 'AnyToolDefinition'] | None" = None, tool_choice: 'ToolChoice | None' = None, input_audio_media_type: 'str | None' = None, output_audio_media_type: 'str | None' = None, input_sample_rate_hz: 'int | None' = None, output_sample_rate_hz: 'int | None' = None, channels: 'int | None' = None, translation_target_language_code: 'str | None' = None, translation_echo_target_language: 'bool | None' = None, turn_detection: 'dict[str, Any] | None' = None, provider_options: 'ProviderOptions | None' = None, metadata: 'dict[str, Any]' = <factory>, auto_response: 'bool' = True) -> None
RealtimeSessionConfig(instructions: 'str | None' = None, voice: 'str | None' = None, tools: "dict[str, 'AnyToolDefinition'] | None" = None, tool_choice: 'ToolChoice | None' = None, input_audio_media_type: 'str | None' = None, output_audio_media_type: 'str | None' = None, input_sample_rate_hz: 'int | None' = None, output_sample_rate_hz: 'int | None' = None, channels: 'int | None' = None, translation_target_language_code: 'str | None' = None, translation_echo_target_language: 'bool | None' = None, turn_detection: 'dict[str, Any] | None' = None, provider_options: 'ProviderOptions | None' = None, metadata: 'dict[str, Any]' = <factory>, auto_response: 'bool' = True)

RealtimeSessionEndedEvent

Experimental · types

from zhivex_ai.experimental import RealtimeSessionEndedEvent
RealtimeSessionEndedEvent(type: "Literal['realtime-end']" = 'realtime-end', reason: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>) -> None
RealtimeSessionEndedEvent(type: "Literal['realtime-end']" = 'realtime-end', reason: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>)

RealtimeSessionStartedEvent

Experimental · types

from zhivex_ai.experimental import RealtimeSessionStartedEvent
RealtimeSessionStartedEvent(type: "Literal['realtime-start']" = 'realtime-start', session_id: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>) -> None
RealtimeSessionStartedEvent(type: "Literal['realtime-start']" = 'realtime-start', session_id: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>)

RealtimeTextDeltaEvent

Experimental · types

from zhivex_ai.experimental import RealtimeTextDeltaEvent
RealtimeTextDeltaEvent(type: "Literal['realtime-text-delta']" = 'realtime-text-delta', text_delta: 'str' = '', item_id: 'str | None' = None, response_id: 'str | None' = None, role: "Literal['assistant']" = 'assistant', provider_metadata: 'dict[str, Any]' = <factory>) -> None
RealtimeTextDeltaEvent(type: "Literal['realtime-text-delta']" = 'realtime-text-delta', text_delta: 'str' = '', item_id: 'str | None' = None, response_id: 'str | None' = None, role: "Literal['assistant']" = 'assistant', provider_metadata: 'dict[str, Any]' = <factory>)

RealtimeTokenResult

Experimental · types

from zhivex_ai.experimental import RealtimeTokenResult
RealtimeTokenResult(value: 'str', expires_at_ms: 'int | None' = None, raw_response: 'Any' = None) -> None
RealtimeTokenResult(value: 'str', expires_at_ms: 'int | None' = None, raw_response: 'Any' = None)

RealtimeToolCallEvent

Experimental · types

from zhivex_ai.experimental import RealtimeToolCallEvent
RealtimeToolCallEvent(type: "Literal['realtime-tool-call']" = 'realtime-tool-call', tool_call: 'ToolCall' = <factory>) -> None
RealtimeToolCallEvent(type: "Literal['realtime-tool-call']" = 'realtime-tool-call', tool_call: 'ToolCall' = <factory>)

RealtimeToolResultEvent

Experimental · types

from zhivex_ai.experimental import RealtimeToolResultEvent
RealtimeToolResultEvent(type: "Literal['realtime-tool-result']" = 'realtime-tool-result', tool_result: 'ToolExecutionResult' = <factory>) -> None
RealtimeToolResultEvent(type: "Literal['realtime-tool-result']" = 'realtime-tool-result', tool_result: 'ToolExecutionResult' = <factory>)

RealtimeTranscriptEvent

Experimental · types

from zhivex_ai.experimental import RealtimeTranscriptEvent
RealtimeTranscriptEvent(type: "Literal['realtime-transcript']" = 'realtime-transcript', text: 'str' = '', role: "Literal['user', 'assistant']" = 'assistant', is_final: 'bool' = False, item_id: 'str | None' = None, response_id: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>) -> None
RealtimeTranscriptEvent(type: "Literal['realtime-transcript']" = 'realtime-transcript', text: 'str' = '', role: "Literal['user', 'assistant']" = 'assistant', is_final: 'bool' = False, item_id: 'str | None' = None, response_id: 'str | None' = None, provider_metadata: 'dict[str, Any]' = <factory>)

create_bedrock

Experimental · provider

from zhivex_ai.experimental import create_bedrock
create_bedrock(*, client: 'BedrockClient | None' = None, region: 'str | None' = None, realtime_connection_factory: 'RealtimeConnectionFactory | None' = None)

native-only provider

create_ollama

Experimental · provider

from zhivex_ai.experimental import create_ollama
create_ollama(*, api_key: 'str | None' = 'ollama', base_url: 'str' = 'http://localhost:11434/v1', fetch: 'Fetcher | None' = None)

compatibility provider

create_openrouter

Experimental · provider

from zhivex_ai.experimental import create_openrouter
create_openrouter(*, api_key: 'str | None' = None, base_url: 'str' = 'https://openrouter.ai/api/v1', fetch: 'Fetcher | None' = None)

native-only provider

open_websocket_connection

Experimental · realtime

from zhivex_ai.experimental import open_websocket_connection
open_websocket_connection(url: 'str', *, headers: 'dict[str, str] | None' = None, options: 'RealtimeConnectOptions | None' = None) -> 'RealtimeConnection'

openai_local_shell_tool

Experimental · provider

from zhivex_ai.experimental import openai_local_shell_tool
openai_local_shell_tool(**extra: 'Any') -> 'HostedToolDefinition'

local shell execution is experimental and must be isolated by applications

openai_shell_environment

Experimental · provider

from zhivex_ai.experimental import openai_shell_environment
openai_shell_environment(*, file_ids: 'list[str] | None' = None, memory_limit: 'str | None' = None, network_policy: 'dict[str, Any] | None' = None, container_id: 'str | None' = None, local_skills: 'list[dict[str, Any]] | None' = None, use_local: 'bool' = False, **extra: 'Any') -> 'HostedToolDefinition'

openai_shell_tool

Experimental · provider

from zhivex_ai.experimental import openai_shell_tool
openai_shell_tool(*, environment: 'HostedToolDefinition | dict[str, Any] | None' = None, **extra: 'Any') -> 'HostedToolDefinition'

shell execution is experimental and must be isolated by applications

stream_live_agent

Experimental · agent

from zhivex_ai.experimental import stream_live_agent
stream_live_agent(*, agent: 'Agent[AgentDepsT, AgentOutputT]', session: 'AgentSession | None' = None, deps: 'AgentDepsT | None' = None, tools: 'ToolSet | ToolRegistry | None' = None, skills: 'SkillSet | SkillRegistry | None' = None, tool_choice: 'str | ToolChoiceName | None' = None, tool_execution: 'ToolExecutionOptions | None' = None, connect_options: 'RealtimeConnectOptions | None' = None, realtime_config: 'RealtimeSessionConfig | None' = None, provider_options: 'dict[str, Any] | None' = None, runtime: 'AgentRuntime | None' = None, registry: 'AgentRegistry | None' = None, observer: 'AgentObserver | None' = None, prompt: 'str | None' = None, messages: 'list[ModelMessage] | None' = None, parent_run_id: 'str | None' = None, idempotency_key: 'str | None' = None, cancellation_token: 'AgentCancellationToken | None' = None, hooks: 'Iterable[AgentHooks] | None' = None, middleware: 'Iterable[AgentMiddleware] | None' = None, stream_buffer_size: 'int | None' = None) -> 'LiveAgentStreamResult[AgentOutputT]'

realtime/live APIs are experimental

Zhivex AI SDKsPortable by default. Native when needed.
Copied