Nemotron 3.5 Lightning 30B

🔥⚡

nemotron-3.5-lightning:30b

nemotron-3.530bBF16ToolsReasoningSuperFast

NVIDIA's Nemotron 3.5 Lightning, a 30B model built for fast, low-cost text generation. Supports tool calling, JSON output and streaming, with a 262K-token context window.

When to choose Nemotron 3.5 Lightning 30B

Nemotron 3.5 Lightning 30B is NVIDIA's open-weight model built for fast, low-cost text generation. It is a mixture-of-experts model: only a small part of its parameters is active for each token, so it starts answering quickly and keeps the price per token low. It works with text only and supports tool calling, JSON output and streaming.

It is a strong fit for:

High-volume text work — classification, tagging, extraction and short summaries.
Chat and assistants — quick, direct replies where latency decides the experience.
Simple tool calls — routing, lookups and well-described function calls in agent loops.
A first, inexpensive pass — try a task here before paying for a larger model.

Reasoning on or off

Reasoning is off by default, so replies come straight away. Turn it on for problems that benefit from step-by-step thinking; reasoning tokens are billed as output tokens. See Reasoning controls.

How it compares

For harder coding, writing and research, move up to Qwen3.8 27B. For agentic coding with image input, compare Ornith 1.5 35B. When the input is very long, use GLM 5.3 Flash or DeepSeek V4.1 Flash for their much larger context window.

Pricing

Input (per 1M tokens)US$0.07 / M Tokens

Output (per 1M tokens)US$0.17 / M Tokens

Cache read (per 1M tokens)US$0.02 / M Tokens

You pay only for the tokens you use. How billing works.

Specifications

Context window256K tokens

Max output256K tokens

InputText

OutputText

ReasoningOff, On (default Off)

Performance

Latency 0.3 s

Throughput 600 tok/s · measured Oct 10, 2026

Latency and throughput: median of the last 10 test requests from our server.

Uptime 100.0 % · last 3 days

Share of our automatic test requests in the last 3 UTC days (today so far and the two days before) that got a complete answer. We send them about every 30 minutes from our server to the suppliers your requests use by default; supplier timeouts and rate limits count as failures. A short outage between two checks can be missed.

Quick start

Call Nemotron 3.5 Lightning 30B through the OpenAI-compatible API at https://a.tokener.ca/v1, or the Anthropic-compatible (Messages) API at https://a.tokener.ca (no /v1), with the model id nemotron-3.5-lightning:30b.

curl https://a.tokener.ca/v1/chat/completions \  -H "Authorization: Bearer $TOKENER_API_KEY" \  -H "Content-Type: application/json" \  -d '{    "model": "nemotron-3.5-lightning:30b",    "messages": [{"role": "user", "content": "Hello!"}],    "stream": true  }'

import osfrom openai import OpenAI client = OpenAI(    base_url="https://a.tokener.ca/v1",    api_key=os.environ["TOKENER_API_KEY"],) stream = client.chat.completions.create(    model="nemotron-3.5-lightning:30b",    messages=[{"role": "user", "content": "Hello!"}],    stream=True,)for chunk in stream:    print(chunk.choices[0].delta.content or "", end="")

import OpenAI from "openai"; const client = new OpenAI({  baseURL: "https://a.tokener.ca/v1",  apiKey: process.env.TOKENER_API_KEY,}); const stream = await client.chat.completions.create({  model: "nemotron-3.5-lightning:30b",  messages: [{ role: "user", content: "Hello!" }],  stream: true,});for await (const chunk of stream) {  process.stdout.write(chunk.choices[0]?.delta?.content ?? "");}

import { createOpenAICompatible } from "@ai-sdk/openai-compatible";import { streamText } from "ai"; const tokener = createOpenAICompatible({  name: "tokener",  baseURL: "https://a.tokener.ca/v1",  apiKey: process.env.TOKENER_API_KEY,}); const result = streamText({  model: tokener.chatModel("nemotron-3.5-lightning:30b"),  messages: [{ role: "user", content: "Hello!" }],}); for await (const text of result.textStream) {  process.stdout.write(text);}

import osfrom anthropic import Anthropic client = Anthropic(    base_url="https://a.tokener.ca",    api_key=os.environ["TOKENER_API_KEY"],) with client.messages.stream(    model="nemotron-3.5-lightning:30b",    max_tokens=4096,    messages=[{"role": "user", "content": [{"type": "text", "text": "Hello!"}]}],) as stream:    for text in stream.text_stream:        print(text, end="")

# Point Claude Code at Tokener, then start it in your project directory.export ANTHROPIC_BASE_URL=https://a.tokener.ca            # no /v1 — the tool appends itexport ANTHROPIC_AUTH_TOKEN=$TOKENER_API_KEY              # your Tokener API keyexport ANTHROPIC_MODEL=nemotron-3.5-lightning:30b                         # the main modelexport ANTHROPIC_DEFAULT_HAIKU_MODEL=nemotron-3.5-lightning:30b           # background calls — a cheaper model from /models also worksexport ANTHROPIC_SMALL_FAST_MODEL=nemotron-3.5-lightning:30bunset ANTHROPIC_API_KEYclaude

Other protocols

The same model, key and price are available on the OpenAI Responses and Anthropic-compatible (Messages) protocols. For the Anthropic SDKs, use the base URL https://a.tokener.ca (without /v1).

OpenAI Responses — guide

curl https://a.tokener.ca/v1/responses \
  -H "Authorization: Bearer $TOKENER_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "nemotron-3.5-lightning:30b",
    "input": [{"role": "user", "content": [{"type": "input_text", "text": "Hello!"}]}],
    "stream": false
  }'

Anthropic-compatible (Messages) — guide

curl https://a.tokener.ca/v1/messages \
  -H "x-api-key: $TOKENER_API_KEY" \
  -H "anthropic-version: 2023-06-01" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "nemotron-3.5-lightning:30b",
    "max_tokens": 4096,
    "messages": [{"role": "user", "content": [{"type": "text", "text": "Hello!"}]}],
    "stream": false
  }'

Questions

How much does Nemotron 3.5 Lightning 30B cost?

Nemotron 3.5 Lightning 30B costs US$0.07 per 1M input tokens and US$0.17 per 1M output tokens. Cached input tokens are US$0.02 per 1M. You pay only for the tokens you use.

What is the context window of Nemotron 3.5 Lightning 30B?

Nemotron 3.5 Lightning 30B has a context window of 262,144 tokens. It can return up to 262,144 output tokens.

Can I control reasoning on Nemotron 3.5 Lightning 30B?

Yes — Nemotron 3.5 Lightning 30B supports these reasoning settings: Off, On. The default is Off. You choose it per request.

Is Nemotron 3.5 Lightning 30B OpenAI-compatible?

Yes — call Nemotron 3.5 Lightning 30B through Tokener's OpenAI-compatible API at https://a.tokener.ca/v1 with the model id nemotron-3.5-lightning:30b. Both OpenAI protocols work: Chat Completions at https://a.tokener.ca/v1/chat/completions and OpenAI Responses at https://a.tokener.ca/v1/responses. Any OpenAI-compatible client or SDK works.

Can I use Nemotron 3.5 Lightning 30B with the Anthropic SDK?

Yes — Nemotron 3.5 Lightning 30B is available on the Anthropic-compatible Messages API at https://a.tokener.ca/v1/messages. Point the Anthropic SDK at the base URL https://a.tokener.ca (without /v1 — the SDK appends it) and use the model id nemotron-3.5-lightning:30b. Which features work varies by model; see the Anthropic SDK guide at /docs/anthropic-sdk.

How do I try Nemotron 3.5 Lightning 30B?

Create an API key, add credits, and call the endpoint with the model id nemotron-3.5-lightning:30b — or try Nemotron 3.5 Lightning 30B in the Tokener Playground.

Try in Playground Get an API key Integration docs All models