Ctrl K
Back to models

NVIDIA

Nemotron 3.5 Lightning

  • Released
  • Open weights
  • OpenMDW-1.1
Compare
Intelligence Index

Reasoning setting

12.9#63 of 83measured models
Measured · Artificial Analysis v4.3.2
API price from
$0.07 in / $0.20 out / 1M tokOpenRouter · CoreWeave · BF16
Context window
1M tok

Performance profile

ReasoningRank #68 of 107
74.3%
GPQA Diamond
CodingRank #78 of 94
32.1%
SciCode
Cost efficiencyRank #22 of 67
138.1pts / $
Intelligence ÷ eval cost

Capabilities

Context window

Modalities

Measured setting

Integration

Reasoning trade-offs

Source (opens in a new tab)
Intelligence Index v4.3.2
Nemotron 3.5 Lightning
Cost per task (USD) · log scale
Nemotron 3.5 LightningReasoning
Intelligence Index
12.9
Cost / task
$0.0931
Evaluation time
1.9m

Pricing & access

API route
OpenRouter · CoreWeave · BF16API
Tier & contextStandard

USD per 1M tokens

Input
$0.07
Output
$0.20
Cached input
$0.04

Run locally

GGUF
Context
Estimated memory

~26 GiB

RAM + VRAM

Estimate assumptions

One text chat at the selected context, with a 16-bit cache and app allowance. Operating system and image processing are extra; optimized caches can use less.

32 GiB RAMwithout GPU offload

Memory budgetOne text chat
Weights
23.5 GiB
Context + state
95.6 MiB16-bit cache
App allowance
2.4 GiB
Download
25.3 GB
Disk space
~33 GB
Available software
Ollama (opens in a new tab)LM Studio (opens in a new tab)Jan (opens in a new tab)KoboldCpp (opens in a new tab)llama.cpp (opens in a new tab)
Get GGUF (opens in a new tab)
Set up in Ollama · Q4_K_M · 8K context

Save the following as Modelfile. The context matches the estimate above.

FROM hf.co/unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF:Q4_K_M
PARAMETER num_ctx 8192

Then run these commands in your terminal:

ollama create llmcurrent-local -f Modelfile
ollama run llmcurrent-local

Ollama import instructions (opens in a new tab) · For LM Studio or Jan, open the app’s model search, enter unsloth/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-GGUF, choose Q4_K_M, and set context to 8,192 tokens in the app’s model settings. Software links above open each app’s setup guide.

vLLM setup (opens in a new tab)

Benchmarks

Full leaderboardSource (opens in a new tab)

Humanity’s Last Exam

Frontier reasoning

Nemotron 3.5 Lightning · Reasoning#90 of 126

  1. 1Claude Opus 5.5Max61.4%
  2. 2Claude Fable 5.1Max59.1%
  3. 3Gemini 4 ArgonHigh57.1%
  5. 87North Mini CodeReasoning11.1%
  6. 88gpt-oss-20bHigh11.0%
  7. 89Hermes 4 - Llama-3.1 405BReasoning10.9%
  8. 90Nemotron 3.5 LightningReasoning10.6%
  9. 91Qwen3 Coder NextStandard10.1%
  10. 92Qwen3.5 4BReasoning9.9%
  11. 93Mistral Small 4Reasoning9.9%

Peers use their best measured result on this benchmark.

Alternatives

Each model at its highest-scoring measured setting · Intelligence Index v4.3.2 and its evaluation cost per task.