NOUMENOpen app
NOUMEN

Meta · Open weights

Llama 3.2 3b Instruct

Added Jan 1, 2024Open weights

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Use in the app

Small model optimized for advanced natural language processing tasks like dialogue generation, reasoning, and summarization

  • PDF input
Input
Text
Output
Text
Discovery tier
Open weights

Pricing

USD per 1M tokens
Input
$0.0306
Output
$0.0493

Reference prices as published by the model provider. API example

Specifications

Context window
131.1K
Max output
8.2K
Average output (7 days)
119 tokens
Longer than 11% of 465 models

Side by side

Llama 3.2 3b Instruct compared

Specifications, prices, speed and benchmark scores next to each other. Benchmarks below show every compared model too.

Compare models

Choose up to 3 models to compare.1 of 3 chosen

Loading models…

Llama 3.2 3b InstructMetaThis modelMuse WonderMetaRemove
ProviderMetaMeta
Discovery tierOpen weightsFrontier
AddedJan 1, 2024Oct 12, 2026
Limits
Context window131.07K1.05MBest
Max output8.19K—
Average output119 tokens1.2K tokens
Pricing, USD per 1M tokens
Input$0.0306Best$4.00
Output$0.0493Best$20.00
Cache read—$0.20
Performance
Throughput160.4 tok/s200.7 tok/sBest
Time to first token0.8 sBest2.9 s
Request success (24h)100.0%Best99.7%
Benchmark indexes
Intelligence Index5.752.4Best
Math Index3.3—
Capabilities
ReasoningNoYes
ToolsNoYes
VisionNoYes
Structured outputNoYes
PDF inputYesYes
Audio inputNoYes
Video inputNoYes
Reasoning effort—minimal, low, medium, high, xhigh, max

Public evaluations

Benchmarks

Independent evaluation results, with how each score ranks among the models in the Noumen catalog.

  • Llama 3.2 3b Instruct
  • Muse Wonder

Intelligence Index

  • Llama 3.2 3b Instruct5.7
  • Muse Wonder52.4

Llama 3.2 3b Instruct: Better than 2% of 326 scored models

3.8Catalog models by score55.1

Indexes

  • Math Index

    Composite of the math evaluations

    • Llama 3.2 3b Instruct: 3.3
    • Muse Wonder: —

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • Llama 3.2 3b Instruct: —
    • Muse Wonder: 68.1%

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • Llama 3.2 3b Instruct: —
    • Muse Wonder: 39.1%

  • Harvey LAB-AAv1.1

    Legal tasks passed in full, with no material hallucination

    • Llama 3.2 3b Instruct: —
    • Muse Wonder: 10.0%

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • Llama 3.2 3b Instruct: —
    • Muse Wonder: 1,659 Elo

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • Llama 3.2 3b Instruct: —
    • Muse Wonder: 1,770 Elo

Document reasoning

  • GDP.pdf

    Professional reasoning over PDF documents, all-pass rate

    • Llama 3.2 3b Instruct: —
    • Muse Wonder: 30.6%

  • AA-LCRv1.1

    Reasoning across long documents

    • Llama 3.2 3b Instruct: 4.3%
    • Muse Wonder: 80.7%Top

  • MLCR-AA

    Long-context reasoning over medical records

    • Llama 3.2 3b Instruct: —
    • Muse Wonder: 22.2%

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • Llama 3.2 3b Instruct: 5.3%
    • Muse Wonder: 50.1%Top

  • GPQA Diamond

    Graduate-level science questions

    • Llama 3.2 3b Instruct: 25.5%
    • Muse Wonder: —

  • MMLU-Pro

    Multi-subject knowledge and reasoning

    • Llama 3.2 3b Instruct: 34.7%
    • Muse Wonder: —

  • IFBench

    Following precise output instructions

    • Llama 3.2 3b Instruct: 26.2%
    • Muse Wonder: —

Coding

  • Terminal-Benchv4.0

    Practical software engineering in a terminal

    • Llama 3.2 3b Instruct: —
    • Muse Wonder: 52.0%

  • SciCode

    Scientific computing in Python

    • Llama 3.2 3b Instruct: —
    • Muse Wonder: 60.3%

  • LiveCodeBench

    Recent competitive programming problems

    • Llama 3.2 3b Instruct: 8.3%
    • Muse Wonder: —

Math

  • AIME 2025

    Competition mathematics, 2025 problems

    • Llama 3.2 3b Instruct: 3.3%
    • Muse Wonder: —

  • AIME

    Competition mathematics

    • Llama 3.2 3b Instruct: 6.7%
    • Muse Wonder: —

  • MATH-500

    Competition math problems

    • Llama 3.2 3b Instruct: 48.9%
    • Muse Wonder: —

Legacy evaluations

  • τ²-Bench Telecom

    Tool use in a customer-support simulation

    • Llama 3.2 3b Instruct: 21.1%
    • Muse Wonder: —

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • Llama 3.2 3b Instruct: 4.3%
    • Muse Wonder: 80.7%Top

Scores from Artificial Analysis, updated Oct 26, 2026. Artificial Analysis

Live routing data

Performance and reliability

Recent averages across requests routed for this model. Your own speed depends on prompt length, reasoning effort and load.

Throughput
160.4 tok/s
Faster than 78% of 251 models
Time to first token
0.8 s
Quicker than 98% of 251 models
Prompt cache hits
4.6%
Request success (24h)
100.0%
Reliable

Cache hits are the share of input tokens recently served from a prompt cache across all requests. They are not a forecast of your own hit rate.

Hourly request success · last 72 hoursLowest hour 100.0%
  • Healthy (98% or more)
  • Degraded (90–98%)
  • Down (under 90%)
  • No requests

Developers

Use it through the API

Noumen’s API is OpenAI-compatible: any Chat Completions client works with a base URL and a key.

Read the quickstart
curl https://www.noumen.xyz/api/v1/chat/completions \
  -H "Authorization: Bearer $NOUMEN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "meta-llama/llama-3.2-3b-instruct",
    "messages": [{ "role": "user", "content": "Explain this stack trace." }]
  }'
import os
from openai import OpenAI

client = OpenAI(base_url="https://www.noumen.xyz/api/v1", api_key=os.environ["NOUMEN_API_KEY"])
reply = client.chat.completions.create(
    model="meta-llama/llama-3.2-3b-instruct",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)

Meta

Other Meta models, starting with the same family.