Skip to content

Llama 3.2 3B Instruct

Llama 3.2 3B Instruct offers capability gains over the 1B, outperforming comparable small models on summarization, instruction following, and tool use.

index.ts
import { streamText } from 'ai'


const result = streamText({
  model: 'meta/llama-3.2-3b',
  prompt: 'Why is the sky blue?'
})
OverviewPlaygroundAboutProvidersThroughputLatencyUptimeStatusSimilarFAQ

Frequently Asked Questions

  • How does Llama 3.2 3B Instruct differ from the 1B beyond just having more parameters?

    Llama 3.2 3B Instruct outperforms Gemma 2 2.6B and Phi 3.5-mini on instruction following, summarization, prompt rewriting, and tool use. On these tasks the 1B only matches (it does not beat) smaller models like Gemma, so the 3B is a clear step up in quality.

  • What is the context window?

    128K tokens, with consistent quality at both short and long context lengths.

  • Does the 3B support tool calling?

    Yes. Tool calling is a trained capability, enabling the 3B to participate in agentic workflows that call external tools or APIs.

  • When should I choose 3B over 8B?

    If cost and throughput are the dominant factors and your task quality requirements are met by the 3B, it will be cheaper and faster to serve at scale. If task complexity demands the 8B's reasoning depth, the cost difference is typically justified.