Skip to content
Sign inGet started

Copy page
View as MarkdownPlain text for this pageOpen in ChatGPTQuestions about this pageOpen in ClaudeQuestions about this page

@standardagents/groq

Groq provider package for Standard Agents using the official groq-sdk

Overview

The @standardagents/groq package provides a Standard Agents provider for Groq's chat completions API. It uses the official groq-sdk, supports streaming, reasoning-capable models, tool calling, and model discovery through Groq's models API.

Key Features
  • Official groq-sdk integration
  • Tool calling and structured outputs
  • Reasoning support for gpt-oss-* and qwen/qwen3-32b
  • Vision input support for Groq multimodal models
  • Built-in pricing for documented Groq models in request logs

Installation

pnpm
pnpm add @standardagents/groq

Quick Start

import { defineModel } from '@standardagents/spec';
import { groq } from '@standardagents/groq';

export default defineModel({
  name: 'groq-fast',
  provider: groq,
  model: 'llama-3.1-8b-instant',
  providerOptions: {
    reasoning_format: 'parsed',
  },
});

Common Models

  • llama-3.1-8b-instant
  • llama-3.3-70b-versatile
  • meta-llama/llama-4-scout-17b-16e-instruct
  • moonshotai/kimi-k2-instruct
  • openai/gpt-oss-120b
  • openai/gpt-oss-20b
  • qwen/qwen3-32b

Provider Options

Use reasoning_format, reasoning_effort, citation_options, service_tier, disable_tool_validation, search_settings, and documents to access Groq-specific behavior.

Environment Setup

.dev.vars
GROQ_API_KEY=...

Streaming usage reads the final Groq metadata frame. Token counts include cached input and reasoning output when Groq reports them.