NOUMENOpen app
NOUMEN

nvidia · Open weights

Nvidia Nemotron 3.5 Lightning Thinking

Added Aug 12, 2026Open weights

Compare models

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

Loading models…

Use in the app

Nvidia Nemotron 3.5 Lightning is an open-weight 30B-A3B hybrid Mamba-Transformer mixture-of-experts model for high-throughput agentic, coding, tool-use, instruction-following, and long-context workloads. This variant enables its reasoning trace.

  • Reasoning
  • Tools
  • Parallel tools
Input
Text
Output
Text
Reasoning effort
none · high
Discovery tier
Open weights

Pricing

USD per 1M tokens
Input
$0.05
Output
$0.20
Cache read
$0.01

Reference prices as published by the model provider. API example

Specifications

Context window
1M
Max output
65.5K
Average output (7 days)
6.1K tokens
Longer than 99% of 468 models

Side by side

Nvidia Nemotron 3.5 Lightning Thinking 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 3.5 Lightning ThinkingnvidiaThis modelDeepSeek R1 0528DeepSeekRemove
ProvidernvidiaDeepSeek
Discovery tierOpen weightsOpen weights
AddedAug 12, 2026May 28, 2025
Limits
Context window1MBest163.84K
Max output65.54KBest32.77K
Average output6.1K tokens1K tokens
Pricing, USD per 1M tokens
Input$0.05Best$0.50
Output$0.20Best$2.15
Cache read$0.01Best$0.35
Performance
Throughput—33.0 tok/s
Time to first token—1.3 s
Request success (24h)—99.9%
Benchmark indexes
Intelligence Index12.913.1Best
Coding Index26.8—
Math Index—76.0
Capabilities
ReasoningYesYes
ToolsYesYes
Parallel toolsYesNo
Reasoning effortnone, high—

Public evaluations

Benchmarks

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

  • Nvidia Nemotron 3.5 Lightning Thinking
  • DeepSeek R1 0528

Intelligence Index

  • Nvidia Nemotron 3.5 Lightning Thinking12.9
  • DeepSeek R1 052813.1

Nvidia Nemotron 3.5 Lightning Thinking: Better than 30% of 326 scored models

3.8Catalog models by score55.1

Indexes

  • Coding Index

    Composite of the coding evaluations

    • Nvidia Nemotron 3.5 Lightning Thinking: 26.8
    • DeepSeek R1 0528: —

  • Math Index

    Composite of the math evaluations

    • Nvidia Nemotron 3.5 Lightning Thinking: —
    • DeepSeek R1 0528: 76.0

Agentic work

  • AutomationBench-AA

    Workflow automation, penalized for guardrail violations

    • Nvidia Nemotron 3.5 Lightning Thinking: 0.8%
    • DeepSeek R1 0528: —

  • AutomationBench-AA tasks completed

    Workflows finished end to end without a guardrail violation

    • Nvidia Nemotron 3.5 Lightning Thinking: 0.0%
    • DeepSeek R1 0528: —

  • AA-Briefcasev1.1

    Agentic knowledge work, rated head to head

    • Nvidia Nemotron 3.5 Lightning Thinking: 506 Elo
    • DeepSeek R1 0528: —

  • GDPval-AAv2.1

    Economically valuable occupational tasks, rated head to head

    • Nvidia Nemotron 3.5 Lightning Thinking: 638 Elo
    • DeepSeek R1 0528: —

Document reasoning

  • GDP.pdf

    Professional reasoning over PDF documents, all-pass rate

    • Nvidia Nemotron 3.5 Lightning Thinking: 3.0%
    • DeepSeek R1 0528: —

  • AA-LCRv1.1

    Reasoning across long documents

    • Nvidia Nemotron 3.5 Lightning Thinking: 60.3%Top
    • DeepSeek R1 0528: 55.7%

  • MLCR-AA

    Long-context reasoning over medical records

    • Nvidia Nemotron 3.5 Lightning Thinking: 2.2%
    • DeepSeek R1 0528: —

Reasoning and knowledge

  • Humanity’s Last Exam

    Expert-level questions across academic fields

    • Nvidia Nemotron 3.5 Lightning Thinking: 10.6%
    • DeepSeek R1 0528: 15.8%Top

  • GPQA Diamond

    Graduate-level science questions

    • Nvidia Nemotron 3.5 Lightning Thinking: 74.3%
    • DeepSeek R1 0528: 81.3%Top

  • MMLU-Pro

    Multi-subject knowledge and reasoning

    • Nvidia Nemotron 3.5 Lightning Thinking: —
    • DeepSeek R1 0528: 84.9%

  • IFBench

    Following precise output instructions

    • Nvidia Nemotron 3.5 Lightning Thinking: —
    • DeepSeek R1 0528: 39.6%

Coding

  • Terminal-Benchv4.0

    Practical software engineering in a terminal

    • Nvidia Nemotron 3.5 Lightning Thinking: 0.5%
    • DeepSeek R1 0528: —

  • SciCode

    Scientific computing in Python

    • Nvidia Nemotron 3.5 Lightning Thinking: 32.1%
    • DeepSeek R1 0528: —

  • LiveCodeBench

    Recent competitive programming problems

    • Nvidia Nemotron 3.5 Lightning Thinking: —
    • DeepSeek R1 0528: 77.0%

Math

  • AIME 2025

    Competition mathematics, 2025 problems

    • Nvidia Nemotron 3.5 Lightning Thinking: —
    • DeepSeek R1 0528: 76.0%

  • AIME

    Competition mathematics

    • Nvidia Nemotron 3.5 Lightning Thinking: —
    • DeepSeek R1 0528: 89.3%

  • MATH-500

    Competition math problems

    • Nvidia Nemotron 3.5 Lightning Thinking: —
    • DeepSeek R1 0528: 98.3%

Legacy evaluations

  • Terminal-Bench Hard

    Earlier terminal task suite

    • Nvidia Nemotron 3.5 Lightning Thinking: —
    • DeepSeek R1 0528: 15.9%

  • τ²-Bench Telecom

    Tool use in a customer-support simulation

    • Nvidia Nemotron 3.5 Lightning Thinking: —
    • DeepSeek R1 0528: 36.5%

  • AA-LCR (unversioned)

    Long-context reasoning, earlier grading

    • Nvidia Nemotron 3.5 Lightning Thinking: 60.3%Top
    • DeepSeek R1 0528: 55.7%

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.

Prompt cache hits
0.0%
About $0.05 per 1M input tokens with caching

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.

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/nemotron-3.5-lightning:thinking",
    "reasoning_effort": "none",
    "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/nemotron-3.5-lightning:thinking",
    messages=[{"role": "user", "content": "Explain this stack trace."}],
)
print(reply.choices[0].message.content)

nvidia

Other nvidia models, starting with the same family.