ModelTokenBudget

CORE · interface · 类型合同

Provider-injected token accounting.

Core never guesses a character-to-token conversion, so token budgets exist only when a host supplies a real counter.

导入

TypeScript

import type { ModelTokenBudget } from '@jangts/agent-core/model';

JavaScript

/** @typedef {import('@jangts/agent-core/model').ModelTokenBudget} ModelTokenBudget */

其他公开入口：@jangts/agent-core、@jangts/agent-core/model、@jangts/agent-core/types、@jangts/agent-core/structs。

定义

interface ModelTokenBudget

字段结构

字段	类型	必需	说明
counterId	string	必需	Identifies the tokenizer and its version; recorded in every measurement.
contextWindowTokens	number	必需	Total context window of the model, in tokens.
safetyMarginTokens	number | undefined	可省略	Additional reserve subtracted together with the output reservation.
countTokens	(input: { messages: readonly ModelMessage[]; functions?: ModelOptions["functions"]; json?: boolean; }) => number	必需	Counts input tokens for the exact request shape that will be sent.

成员目录

成员	用途
countTokens	Counts input tokens for the exact request shape that will be sent.

成员详情

countTokens

Counts input tokens for the exact request shape that will be sent.

ModelTokenBudget.countTokens(input: { messages: readonly ModelMessage[]; functions?: ModelOptions["functions"]; json?: boolean; }): number

Counts input tokens for the exact request shape that will be sent.

参数	类型	传递	说明
input	{ messages: readonly ModelMessage[]; functions?: ModelOptions["functions"]; json?: boolean; }	必需	—

input 的字段

字段	类型	必需	说明
messages	readonly ModelMessage[]	必需	见类型定义及调用场景。
functions	readonly ModelFunction[] | undefined	可省略	见类型定义及调用场景。
json	boolean | undefined	可省略	见类型定义及调用场景。

返回： number。

使用示例

• ModelRunner：输入可以省写，输出合同保持一致：用真实 Provider 端口执行一次模型交互，返回 value 和本次 invocation。
• 一次绑定，单次多模型与跨次切换：构造时提供稳定模型配置；请求用配置 ID 选择，不反复注入 Provider。
• 模型策略：显式选择与推荐预算：从已登记的模型和推理配置中选择，以本地预算档位补全 Core 设置，并保留逐字段精确覆盖。
• 装饰器：处理类与已有实例方法：类装饰代办定义实例化，方法装饰绑定原对象；输出仍走原 define、构造和 bind。

同一输出的两种输入

入门 · Node.js

TypeScript

import { ModelRunner } from '@jangts/agent-core/model';
const provider = {
  async *stream(messages: readonly { content: string }[]) {
    yield '收到：' + messages.at(-1)!.content;
  },
};
const runner = new ModelRunner({ provider });
const first = await runner.invoke('你好');
const second = await runner.invoke({ input: '你好' });
console.log(first.value, second.value);
console.log(first.invocation.invocationId);

JavaScript

import { ModelRunner } from '@jangts/agent-core/model';
const provider = {
    async *stream(messages) {
        yield '收到：' + messages.at(-1).content;
    },
};
const runner = new ModelRunner({ provider });
const first = await runner.invoke('你好');
const second = await runner.invoke({ input: '你好' });
console.log(first.value, second.value);
console.log(first.invocation.invocationId);

增加预算与请求取消

配置 · Node.js

TypeScript

import { ModelRunner } from '@jangts/agent-core/model';
const provider = {
  async *stream(messages: readonly { content: string }[]) {
    yield '收到：' + messages.at(-1)!.content;
  },
};
const runner = new ModelRunner({ provider, maxCalls: 2, maxInputChars: 4000,
  maxOutputChars: 2000, callTimeoutMs: 15000,
  execution: { signal: AbortSignal.timeout(20000) },
});
const result = await runner.invoke({ input: '你好', stageId: 'reply', maxOutputTokens: 600 });
console.log(result.value, runner.remainingCalls);

JavaScript

import { ModelRunner } from '@jangts/agent-core/model';
const provider = {
    async *stream(messages) {
        yield '收到：' + messages.at(-1).content;
    },
};
const runner = new ModelRunner({ provider, maxCalls: 2, maxInputChars: 4000,
    maxOutputChars: 2000, callTimeoutMs: 15000,
    execution: { signal: AbortSignal.timeout(20000) },
});
const result = await runner.invoke({ input: '你好', stageId: 'reply', maxOutputTokens: 600 });
console.log(result.value, runner.remainingCalls);

相关策略源码

• Core 的基础预算缺省值
• 推荐预算、同合同补全和覆盖顺序

相关定义

• ModelMessage
• ModelOptions