Nemotron 3 token counter
Updated
In short: This counter runs NVIDIA's official Nemotron 3 tokenizer.json in your browser, pinned to the same commit as our tokenizer dataset. It has 131,072 token IDs and is a 17 MB download. English prose takes about 130 tokens per 100 words, Modern Standard Arabic about 176.
tokens
| Model | Tokens | Chars per token | As input | As output |
|---|---|---|---|---|
| NVIDIA Nemotron 3.5 Lightning 30B A3B Nemotron 3 tokenizer (Nemotron 3 Ultra) Price: NVIDIA Nemotron 3.5 Lightning 30B A3B at Crusoe Cloud, cheapest listed input price (not BroadNet) | … | … | … | … |
| DeepSeek V4.1 Flash DeepSeek V4.1 Flash tokenizer | … | … | … | … |
| GLM 5.3 GLM 5.3 tokenizer | … | … | … | … |
| Qwen3.8 27B Qwen3.8 27B tokenizer | … | … | … | … |
| Gemma 4 VL 26B Gemma 4 tokenizer (Google base model) | … | … | … | … |
| GPT (o200k) OpenAI o200k_base | … | … | … | … |
Cost = tokens × the model's published price per 1M input or output tokens. GPT is shown for reference only (no BroadNet price). Nemotron 3 models are not sold by BroadNet: its row uses the cheapest standard input price we hold for a model with that tokenizer, read from the provider's own pricing page. Sources and check dates are in the real cost by language dataset. The chat-format option adds template tokens only for the DeepSeek, GLM, Qwen, Gemma and GPT rows; Nemotron 3 row counts your text alone. Measured ratios for every tokenizer are in the tokenizer efficiency dataset. Rows marked ≈ are estimates from measured characters per token until you load that tokenizer; loading downloads its tokenizer.json from Hugging Face once and keeps it in your browser cache.
Other counters: DeepSeek · GLM · Qwen · Gemma · GPT · Llama 3 · Llama 4 · Kimi · Mistral Small · Ministral 3 · gpt-oss · Phi-4 · Gemma 3 · MiniMax · Qwen3 · all models. BroadNet prices synced 2026-11-14.
Report a wrong price or number (email to info@broadnet.ai, with this page's link and your inputs filled in).
How it works
- Paste text into the box or pick a sample. The Nemotron 3 tokenizer (Nemotron 3 Ultra) is preselected and downloads once (17 MB) when you start typing.
- The text appears split into tokens, a colour per token; tick "Show token IDs" to see the numbers.
- Compare the Nemotron 3 count with the other tokenizers in the table; load any of them for an exact count.
- Copy the link or download a CSV of the counts and costs.
About the Nemotron 3 tokenizer
The Nemotron 3 tokenizer is a byte-level BPE. It has 131,072 token IDs, 1,000 of them added or special tokens such as chat-format markers. Text is split into words, punctuation and numbers first, then merged byte by byte. Numbers are split into single digits, so 2026 is four tokens.
On the open texts of our tokenizer efficiency dataset it needs 1,302 tokens per 1,000 English words (rank 14 of the 16 tokenizers we measured, 1 = fewest), 2,096 per 1,000 words of Python (rank 12 of 16) and 1,760 per 1,000 words of Modern Standard Arabic (rank 2 of 16). Its best showing here is Arabic MSA (rank 2 of 16); its weakest is Chinese (rank 15 of 16).
4 models in our price database use this tokenizer: NVIDIA Nemotron 3 Ultra, NVIDIA Nemotron 3 Super 120B A12B, NVIDIA Nemotron 3.5 Lightning 30B A3B and NVIDIA Nemotron 3 Nano 30B A3B. We hold 20 standard prices for them from 13 providers. The cheapest input price is $0.05 per 1M tokens, for NVIDIA Nemotron 3.5 Lightning 30B A3B on Crusoe Cloud (checked 2026-11-15). At that price 1,000 English words cost $0.000065 to send. Nemotron 3 models are not sold by BroadNet. That is less than BroadNet's cheapest chat model, Gemma 4 VL 26B, at $0.00013 for the same text.
| Tokenizer | Nemotron 3 tokenizer (Nemotron 3 Ultra), a byte-level BPE |
|---|---|
| Source | nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16 (commit 77df655) |
| Token IDs | 131,072, special tokens included |
| Models using it | 4: NVIDIA Nemotron 3 Ultra, NVIDIA Nemotron 3 Super 120B A12B, NVIDIA Nemotron 3.5 Lightning 30B A3B and NVIDIA Nemotron 3 Nano 30B A3B |
| Cheapest input price | $0.05 per 1M tokens (NVIDIA Nemotron 3.5 Lightning 30B A3B, Crusoe Cloud, checked 2026-11-15) |
| English prose | 130 tokens per 100 words (rank 14 of 16) |
| Arabic (MSA) | 176 tokens per 100 words (rank 2 of 16) |
| Python code | 210 tokens per 100 words (rank 12 of 16) |
| Download | 17 MB, once; then cached in your browser |
Tokens per 100 words measured on the open texts of our dataset with all 16 tokenizers (2026-11-15); rank 1 = fewest tokens on the same texts. Model mapping checked 2026-11-15. Third-party prices are standard-tier prices from each provider's own page; BroadNet prices come from its pricing.json.
How much does a Nemotron 3 prompt cost?
input cost = tokens × input price per 1M ÷ 1,000,000
output cost = tokens × output price per 1M ÷ 1,000,000Worked example: "English Wikipedia: Water" in our dataset is 274 words. The Nemotron 3 tokenizer (Nemotron 3 Ultra) splits it into 361 tokens. Sent to NVIDIA Nemotron 3.5 Lightning 30B A3B on Crusoe Cloud at $0.05 per 1M input tokens it costs $0.000018; the same number of tokens written as output at $0.20 per 1M costs $0.000072.
Questions
Which tokenizer does Nemotron 3 use?
A byte-level BPE with 131,072 token IDs, published as tokenizer.json in nvidia/NVIDIA-Nemotron-3-Ultra-550B-A55B-BF16. This page loads that file pinned to commit 77df655, the commit our dataset measured.
Which models use the Nemotron 3 tokenizer?
NVIDIA Nemotron 3 Ultra, NVIDIA Nemotron 3 Super 120B A12B, NVIDIA Nemotron 3.5 Lightning 30B A3B and NVIDIA Nemotron 3 Nano 30B A3B. Each was checked against the model's own Hugging Face repo on 2026-11-15: same tokenizer file, or the same vocabulary and rules with identical counts on every dataset text. Not covered: NVIDIA Nemotron 3.5 Super VL (different tokenizer file).
What is the cheapest Nemotron 3 API price per token?
Of the 20 standard prices we hold for Nemotron 3 models, the lowest input price is $0.05 per 1M tokens (output $0.20), for NVIDIA Nemotron 3.5 Lightning 30B A3B on Crusoe Cloud, read from the provider's pricing page on 2026-11-15. Promotional prices are not ranked. Prices differ by model size and provider: each model's price page lists every provider we hold.
How many tokens does Arabic take with Nemotron 3?
Modern Standard Arabic sentences take 1.25× the tokens of their English translations with this tokenizer (rank 2 of 16 on the same texts). Hindi takes 1.80× and Chinese 1.42×. Measured on parallel sentences in our open dataset (2026-11-15). Paste your own text above, or use the Arabic token counter.
Is the Nemotron 3 tokenizer the same as Ministral 3?
They gave identical counts on all 52 texts of our dataset, but they are separate tokenizers with different special tokens, so text that contains control markers can count differently. Load both in the table above to compare your own text. Ministral 3 has its own counter.
Does the count include chat-template tokens?
No. For Nemotron 3 this page counts your text only, as our dataset does. A chat request also adds the chat template's role markers and special tokens, so the billed prompt is a little longer than your text. Read usage.prompt_tokens on a real response to see the exact figure.