Enum TokenEventV2 

pub enum TokenEventV2 {
    Text(String),
    Thinking(String),
    ToolUse {
        tool_call_id: ToolCallId,
        name: String,
        input: ToolUseInput,
    },
    Done {
        stop_reason: StopReasonV2,
        usage: UsageV2,
    },
}

One event in a v2 generation stream — typed-content-block surface per ADR 0015.

v2 separates user-visible text (Text) from reasoning trace (Thinking) and emits complete tool-call requests (ToolUse) as their own variant rather than raw tokens. Backends that don’t distinguish thinking content (any non-Gemma-4 backend) emit only Text events.

Variants

Text(String)

Incremental user-visible text.

Thinking(String)

Incremental reasoning trace (Gemma 4 <|think|> content).

ToolUse

Complete tool-call request emitted by the model.

Fields

tool_call_id: ToolCallId

Identifier paired with the consumer’s eventual ToolResult.

name: String

Tool name from the request’s tools[] table.

input: ToolUseInput

JSON arguments emitted by the model.

Done

Final event for a successful generation.

Fields

stop_reason: StopReasonV2

Reason generation stopped.

usage: UsageV2

Token-count usage.

Trait Implementations

impl Clone for TokenEventV2

fn clone(&self) -> TokenEventV2

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for TokenEventV2

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl PartialEq for TokenEventV2

fn eq(&self, other: &TokenEventV2) -> bool

Tests for self and other values to be equal, and is used by ==.

fn ne(&self, other: &Rhs) -> bool

Tests for !=. The default implementation is almost always sufficient, and should not be overridden without very good reason.

impl StructuralPartialEq for TokenEventV2