NOUMENOpen app
NOUMEN

Meta · Open weights

Llama 3.3 70B

Added Jul 3, 2025Open weights

Compare models

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

Loading models…

Use in the app

Meta's Llama 3.3 with 70B parameters. TEE-backed inference with provider attestation support.

Input
Text
Output
Text
Discovery tier
Open weights

Pricing

USD per 1M tokens
Input
$1.75
Output
$2.75
Cache read
$1.75

Reference prices as published by the model provider. API example

Specifications

Context window
128K
Max output
16.4K
Average output (7 days)
339 tokens
Longer than 36% of 465 models

Side by side

Llama 3.3 70B 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 3.3 70BMetaThis modelMuse Spark 1.2MetaRemove
ProviderMetaMeta
Discovery tierOpen weightsFrontier
AddedJul 3, 2025Aug 5, 2026
Limits
Context window128K1MBest
Max output16.38K65.54KBest
Average output339 tokens172 tokens
Pricing, USD per 1M tokens
Input$1.75$1.25Best
Output$2.75Best$4.25
Cache read$1.75$0.15Best
Benchmark indexes
Intelligence Index7.739.6Best
Coding Index11.972.2Best
Math Index7.7—
Capabilities
ReasoningNoYes
ToolsNoYes
VisionNoYes
Structured outputNoYes
PDF inputNoYes
Audio inputNoYes
Video inputNoYes
Parallel toolsNoYes
Reasoning effort—minimal, low, medium, high, xhigh

Public evaluations

Benchmarks

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

  • Llama 3.3 70B
  • Muse Spark 1.2

Intelligence Index

  • Llama 3.3 70B7.7
  • Muse Spark 1.239.6

Llama 3.3 70B: Better than 11% of 326 scored models

3.8Catalog models by score55.1

Indexes

  • Coding Index

    Composite of the coding evaluations

    • Llama 3.3 70B: 11.9
    • Muse Spark 1.2: 72.2Top

  • Math Index

    Composite of the math evaluations

    • Llama 3.3 70B: 7.7
    • Muse Spark 1.2: —

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • Llama 3.3 70B: —
    • Muse Spark 1.2: 40.6%

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • Llama 3.3 70B: —
    • Muse Spark 1.2: 15.7%

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • Llama 3.3 70B: —
    • Muse Spark 1.2: 1,327 Elo

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • Llama 3.3 70B: —
    • Muse Spark 1.2: 1,475 Elo

Document reasoning

  • GDP.pdf

    Professional reasoning over PDF documents, all-pass rate

    • Llama 3.3 70B: —
    • Muse Spark 1.2: 17.4%

  • AA-LCRv1.1

    Reasoning across long documents

    • Llama 3.3 70B: 15.7%
    • Muse Spark 1.2: 79.0%Top

  • MLCR-AA

    Long-context reasoning over medical records

    • Llama 3.3 70B: —
    • Muse Spark 1.2: 7.2%

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • Llama 3.3 70B: 3.6%
    • Muse Spark 1.2: 45.5%Top

  • GPQA Diamond

    Graduate-level science questions

    • Llama 3.3 70B: 49.8%
    • Muse Spark 1.2: 90.4%Top

  • MMLU-Pro

    Multi-subject knowledge and reasoning

    • Llama 3.3 70B: 71.3%
    • Muse Spark 1.2: —

  • IFBench

    Following precise output instructions

    • Llama 3.3 70B: 47.1%
    • Muse Spark 1.2: —

Coding

  • Terminal-Benchv4.0

    Practical software engineering in a terminal

    • Llama 3.3 70B: —
    • Muse Spark 1.2: 7.1%

  • SciCode

    Scientific computing in Python

    • Llama 3.3 70B: —
    • Muse Spark 1.2: 57.4%

  • LiveCodeBench

    Recent competitive programming problems

    • Llama 3.3 70B: 28.8%
    • Muse Spark 1.2: —

Math

  • AIME 2025

    Competition mathematics, 2025 problems

    • Llama 3.3 70B: 7.7%
    • Muse Spark 1.2: —

  • AIME

    Competition mathematics

    • Llama 3.3 70B: 30.0%
    • Muse Spark 1.2: —

  • MATH-500

    Competition math problems

    • Llama 3.3 70B: 77.3%
    • Muse Spark 1.2: —

Legacy evaluations

  • Terminal-Bench Hard

    Earlier terminal task suite

    • Llama 3.3 70B: 3.0%
    • Muse Spark 1.2: —

  • τ²-Bench Telecom

    Tool use in a customer-support simulation

    • Llama 3.3 70B: 26.6%
    • Muse Spark 1.2: —

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • Llama 3.3 70B: 15.7%
    • Muse Spark 1.2: 79.0%Top

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

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": "TEE/llama3-3-70b",
    "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="TEE/llama3-3-70b",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)

Meta

Other Meta models, starting with the same family.