llama-3.3-70b-versatile
Virgin bootstrap template for Claude Code system with Next.js 15, React 19, TypeScript 5, Prisma, Clerk, Deepgram, Groq, and Llama 3.3 70B
- Category: discovered
- Tags: github, community
- Version: 1.0.0
- Author: walid
- Downloads: 0
- Updated: 2025-11-22
Download the latest version (v1.0.0) from this page.
SKILL.md
# Llama 3.3 70B Versatile Model Skill This skill provides patterns and best practices for using Meta's Llama 3.3 70B Versatile model via Groq API for ultra-fast LLM inference. ## What This Skill Covers - **Prompt Engineering** - Special token format and best practices - **Chat Completions** - Single and multi-turn conversations - **Streaming** - Real-time token streaming with SSE - **JSON Mode** - Structured outputs and data extraction - **Tool Calling** - Function calling capabilities - **Long Context** - 128K token context window usage - **Performance** - Model selection (versatile vs specdec) - **Integration** - Groq API, Vercel AI SDK, LangChain ## When to Use This Skill Use this skill when working on: - Chatbots and conversational AI - Code generation and review - Content summarization and analysis - Structured data extraction (JSON) - Long-form document processing - Multi-turn reasoning tasks - Real-time streaming applications ## Key Features - **High Performance** - 276 tokens/sec (fastest provider on Groq) - **Large Context** - 128K tokens for long documents - **Instruction Following** - 92.1 on IFEval (beats GPT-4) - **Coding Excellence** - 88.4 on HumanEval - **Multilingual** - 8 languages supported - **Special Tokens** - Proper prompt formatting required - **JSON Mode** - Native structured output support ## Related Skills - **groq** - For Groq API patterns and best practices - **nextjs** - For API route integration - **react** - For client-side streaming UI - **typescript** - For type-safe implementation ## Documentation See `references/external/llama-3.3-70b-versatile-patterns.md` for comprehensive patterns, examples, and troubleshooting. --- **Skill Version:** 1.0.0 **Last Verified:** November 12, 2025 **Next Review:** December 2025