NVIDIA Nemotron 3.5 Lightning
nvidia/nemotron-3.5-lightningNVIDIA’s open-weight text model for reasoning and tool use.
262K tokens of context — about 196,608 words, on a scale that ends at 2M.
What it can do
- Reasoning
- Tool use
- Open weights
- Takes text
- Gives text
Released 11 Aug 2026 · checked 3 Oct 2026.
About it
Nemotron 3.5 Lightning is NVIDIA’s open-weight text model, with 3 billion active parameters out of 30 billion total. It supports reasoning and tools, and is suited to high-throughput agent tasks and specialized work. Its context window is 262,144 tokens. Pricing is $0.06 per million input tokens and $0.16 per million output tokens.
Review
In a line
Nemotron 3.5 Lightning is an open-weight text model from NVIDIA. It can reason and use tools.
What it's good at
The model is suited to high-volume agent tasks and specialized work. It has 3 billion active parameters within a 30-billion-parameter model.
Price in plain terms
Input costs $0.06 per million tokens, and output costs $0.16 per million tokens. Reading a page of about 500 words costs under 0.01 cents. One chat turn with about 1,000 input tokens and 500 output tokens costs about 0.014 cents; 1,000 such turns cost about $0.14.
Context and limits
It can take in up to 262,144 tokens at once and write up to 32,768 tokens in one answer. It handles text input and produces text output.
Ways to use it
The model has open weights, which means its model files are available for use. Its model ID is nvidia/nemotron-3.5-lightning.
Verdict
Consider it if you need an open-weight text model with reasoning and tool support for high-volume agent tasks or specialized work. Its low stated token prices may suit readers comparing usage costs. Choose another option if you need image input or output; this model is text-only.
Verdict
Nemotron 3.5 Lightning is worth considering for people building text-based agent workflows who value open weights, tool use, and a large context window. Its stated costs are low for the example page and chat workloads. It is not a fit for image tasks, since it accepts and produces text only. The facts describe its capabilities and prices, but do not name consumer apps or a particular provider, so readers looking for a ready-made chat service may need another choice.
- Open weights are available.
- Supports reasoning and tools.
- 262,144-token context window.
- Input and output prices are stated per million tokens.
- Text model suited to high-throughput agent tasks.
- Text input and output only.
- No consumer app or chat interface is named.
- Max answer length is 32,768 tokens.
- Pick it ifText-based agent tasks with tool use; Specialized workloads needing a large context window
- Skip it ifYou need a model that handles images.; You want a named consumer app or chat interface.
Pricing
Input is $0.06 per million tokens and output is $0.16 per million tokens. The examples put a page at under 0.01 cents, one chat turn at about 0.014 cents, and 1,000 turns at about $0.14.
Questions
Who makes Nemotron 3.5 Lightning?
NVIDIA makes it.
Is it open source?
It has open weights, meaning the model files are available for use.
Can it use tools?
Yes. It supports tools and reasoning.
How much text can it take in at once?
Its context window is 262,144 tokens.
How much can it write in one answer?
Up to 32,768 tokens.
How much does it cost?
Input is $0.06 per million tokens and output is $0.16 per million tokens. One example chat turn costs about 0.014 cents.
Can it work with images?
No. It takes text input and produces text output.
On these boards
More from NVIDIA
- SwitchyardVaries in
- Nemotron 3.5 Lightning (free)Free in
- Nemotron 3 Ultra$0.50 in
- Nemotron 3 Ultra (free)Free in
- Nemotron 3.5 Content Safety$0.20 in
- Nemotron 3.5 Content Safety (free)Free in
- Nemotron 3 Nano Omni (free)Free in
- Nemotron 3 Super$0.080 in
- Nemotron 3 Super (free)Free in
- Nemotron 3 Nano 30B A3B$0.050 in