@standardagents/cerebras

Cerebras provider for Standard Agents.

This package wraps Cerebras' OpenAI-compatible Chat Completions API behind the Standard Agents provider interface. It supports:

• streaming and non-streaming text generation
• local function tool calling on supported models
• JSON and structured output modes
• Cerebras reasoning controls
• usage normalization and provider-owned cost calculation for documented models

Install

npm install @standardagents/cerebras @standardagents/spec

Usage

import { cerebras } from '@standardagents/cerebras';

const provider = cerebras({
  apiKey: process.env.CEREBRAS_API_KEY!,
});

const result = await provider.generate({
  model: 'zai-glm-4.7',
  messages: [{ role: 'user', content: 'Give me three concise launch names.' }],
});

console.log(result.content);
console.log(result.usage?.cost);

Factory

The package exports:

• cerebras(config) - provider factory
• CerebrasProvider - provider class
• cerebrasProviderOptions - Zod schema for provider-specific options

Pricing Helpers

Known Cerebras model pricing is available through both the package root and the pure @standardagents/cerebras/pricing subpath:

import { calculateCerebrasUsageCost } from '@standardagents/cerebras/pricing';

const cost = calculateCerebrasUsageCost('zai-glm-4.7', {
  promptTokens: 6298,
  completionTokens: 284,
});

For known models, generated and streamed responses attach the same value to usage.cost. Custom or private Cerebras models should still define explicit model pricing in AgentBuilder because unknown provider prices are not guessed.