NOUMENOpen app
NOUMEN

nvidia · Open weights

Nvidia Nemotron 70b

Added Apr 15, 2025Open weights

Compare models

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

Loading models…

Use in the app

Nvidia's latest Llama fine-tune optimized for instruction following. Early results hints that it might outperform models such as GPT-4o and Claude 3.5 Sonnet.

Input
Text
Output
Text
Discovery tier
Open weights

Pricing

USD per 1M tokens
Input
$0.36
Output
$0.41

Reference prices as published by the model provider. API example

Specifications

Context window
16.4K
Max output
8.2K

Side by side

Nvidia Nemotron 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…

Nvidia Nemotron 70bnvidiaThis modelNvidia Nemotron 3.5 Super VLnvidiaRemove
Providernvidianvidia
Discovery tierOpen weightsOpen weights
AddedApr 15, 2025Oct 15, 2026
Limits
Context window16.38K262.14KBest
Max output8.19K16.38KBest
Average output—1.1K tokens
Pricing, USD per 1M tokens
Input$0.36$0.15Best
Output$0.41Best$0.50
Cache read—$0.02
Performance
Throughput—621.5 tok/s
Time to first token—1.7 s
Request success (24h)22.9%98.3%Best
Benchmark indexes
Intelligence Index6.926.5Best
Math Index11.0—
Capabilities
ReasoningNoYes
ToolsNoYes
VisionNoYes
Structured outputNoYes
Video inputNoYes
Reasoning effort—none, medium, high

Public evaluations

Benchmarks

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

  • Nvidia Nemotron 70b
  • Nvidia Nemotron 3.5 Super VL

Intelligence Index

  • Nvidia Nemotron 70b6.9
  • Nvidia Nemotron 3.5 Super VL26.5

Nvidia Nemotron 70b: Better than 8% of 326 scored models

3.8Catalog models by score55.1

Indexes

  • Math Index

    Composite of the math evaluations

    • Nvidia Nemotron 70b: 11.0
    • Nvidia Nemotron 3.5 Super VL: —

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • Nvidia Nemotron 70b: —
    • Nvidia Nemotron 3.5 Super VL: 14.8%

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • Nvidia Nemotron 70b: —
    • Nvidia Nemotron 3.5 Super VL: 2.0%

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • Nvidia Nemotron 70b: —
    • Nvidia Nemotron 3.5 Super VL: 999 Elo

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • Nvidia Nemotron 70b: —
    • Nvidia Nemotron 3.5 Super VL: 1,162 Elo

Document reasoning

  • GDP.pdf

    Professional reasoning over PDF documents, all-pass rate

    • Nvidia Nemotron 70b: —
    • Nvidia Nemotron 3.5 Super VL: 8.0%

  • AA-LCRv1.1

    Reasoning across long documents

    • Nvidia Nemotron 70b: 8.3%
    • Nvidia Nemotron 3.5 Super VL: 69.0%Top

  • MLCR-AA

    Long-context reasoning over medical records

    • Nvidia Nemotron 70b: —
    • Nvidia Nemotron 3.5 Super VL: 13.3%

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • Nvidia Nemotron 70b: 4.2%
    • Nvidia Nemotron 3.5 Super VL: 30.7%Top

  • GPQA Diamond

    Graduate-level science questions

    • Nvidia Nemotron 70b: 46.5%
    • Nvidia Nemotron 3.5 Super VL: —

  • MMLU-Pro

    Multi-subject knowledge and reasoning

    • Nvidia Nemotron 70b: 69.0%
    • Nvidia Nemotron 3.5 Super VL: —

  • IFBench

    Following precise output instructions

    • Nvidia Nemotron 70b: 30.7%
    • Nvidia Nemotron 3.5 Super VL: —

Coding

  • Terminal-Benchv4.0

    Practical software engineering in a terminal

    • Nvidia Nemotron 70b: —
    • Nvidia Nemotron 3.5 Super VL: 0.0%

  • SciCode

    Scientific computing in Python

    • Nvidia Nemotron 70b: —
    • Nvidia Nemotron 3.5 Super VL: 47.8%

  • LiveCodeBench

    Recent competitive programming problems

    • Nvidia Nemotron 70b: 16.9%
    • Nvidia Nemotron 3.5 Super VL: —

Math

  • AIME 2025

    Competition mathematics, 2025 problems

    • Nvidia Nemotron 70b: 11.0%
    • Nvidia Nemotron 3.5 Super VL: —

  • AIME

    Competition mathematics

    • Nvidia Nemotron 70b: 24.7%
    • Nvidia Nemotron 3.5 Super VL: —

  • MATH-500

    Competition math problems

    • Nvidia Nemotron 70b: 73.3%
    • Nvidia Nemotron 3.5 Super VL: —

Legacy evaluations

  • Terminal-Bench Hard

    Earlier terminal task suite

    • Nvidia Nemotron 70b: 4.5%
    • Nvidia Nemotron 3.5 Super VL: —

  • τ²-Bench Telecom

    Tool use in a customer-support simulation

    • Nvidia Nemotron 70b: 23.1%
    • Nvidia Nemotron 3.5 Super VL: —

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • Nvidia Nemotron 70b: 8.3%
    • Nvidia Nemotron 3.5 Super VL: 69.0%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.

Request success (24h)
22.9%

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": "nvidia/Llama-3.1-Nemotron-70B-Instruct-HF",
    "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="nvidia/Llama-3.1-Nemotron-70B-Instruct-HF",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)

nvidia

Other nvidia models, starting with the same family.