NVIDIAopen weightsNemotron 3 Ultra pricing
$0.60 per 1M input tokens and $2.40 per 1M output tokens. In real work that is 0.33¢ per chat answer and 1.3¢ to summarize a 20-page PDF.
What Nemotron 3 Ultra costs per task
Token counts estimated from a measured model of the same lab (Mistral Large 4), rescaled by tokenizer. Reasoning tokens are billed as output.
| Task | Cost | Tokens used | Price rank | Basis |
|---|---|---|---|---|
| Answer a chat question | 0.33¢$3.31 / 1,000 | 39 in · 516 out · 854 reasoning | 202 of 336 | estimated |
| Reply to a customer email | 0.86¢$8.64 / 1,000 | 251 in · 592 out · 2,946 reasoning | 300 of 336 | estimated |
| Write a 1,500-word blog post | 1.9¢$18.62 / 1,000 | 54 in · 2,039 out · 5,705 reasoning | 234 of 336 | estimated |
| Write 10 product descriptions | 0.75¢$7.53 / 1,000 | 281 in · 839 out · 2,230 reasoning | 184 of 336 | estimated |
| Translate 1,000 words | 3.0¢$29.88 / 1,000 | 1,333 in · 1,301 out · 10,816 reasoning | 273 of 336 | estimated |
| Proofread one page (500 words) | 1.0¢$10.35 / 1,000 | 722 in · 849 out · 3,281 reasoning | 266 of 336 | estimated |
| Summarize a 20-page PDF | 1.3¢$13.13 / 1,000 | 13,217 in · 149 out · 2,019 reasoning | 213 of 336 | estimated |
| Extract data from an invoice | 0.66¢$6.57 / 1,000 | 369 in · 860 out · 1,784 reasoning | 254 of 336 | estimated |
| Analyze a 200-row spreadsheet | 2.8¢$28.38 / 1,000 | 4,718 in · 4,473 out · 6,174 reasoning | 188 of 336 | estimated |
| Answer from 5 retrieved documents (RAG) | 0.99¢$9.89 / 1,000 | 3,728 in · 207 out · 2,981 reasoning | 234 of 336 | estimated |
| Classify 100 support tickets | 1.8¢$18.48 / 1,000 | 1,259 in · 822 out · 6,565 reasoning | 254 of 336 | estimated |
| Review 500 lines of code | 2.5¢$24.59 / 1,000 | 4,603 in · 1,575 out · 7,522 reasoning | 217 of 336 | estimated |
| Write a SQL query | 0.76¢$7.64 / 1,000 | 119 in · 505 out · 2,647 reasoning | 260 of 336 | estimated |
| Run a 50-step coding agent | 67¢$667 / 1,000 | 2,277,775 in · 15,270 out · 132,350 reasoning | 194 of 336 | modeled |
Price at every provider
List prices per 1M tokens (input / output) at each host serving Nemotron 3 Ultra through OpenRouter today. Hosts can differ in quantization, context and speed.
| Provider | Input / output | Cached input | Context | Uptime |
|---|---|---|---|---|
 DeepInfrafp4 | $0.40 / $1.76 | $0.08 | 262K | |
DeepInfrafp4 | $0.50 / $2.20 | $0.10 | 262K | 100% |
| B BaseTenfp4 | $0.60 / $2.40 | $0.12 | 202K | 100% |
| B BaseTenfp4 | $0.60 / $2.40 | $0.12 | 202K | 100% |
| V Venicefp8 | $0.62 / $3.12 | $0.19 | 256K | 100% |
Price history
First seen in our price history on Jun 6, 2026 at $0.50 input / $2.50 output. Since then the output price moved from $2.50 to $2.40 per 1M tokens (6 changes).
FAQ
How much does Nemotron 3 Ultra cost?
Nemotron 3 Ultra costs $0.60 per 1 million input tokens and $2.40 per 1 million output tokens, with cached input at $0.12 (list price on Oct 18, 2026). In practice one chat answer costs about 0.33¢ and summarizing a 20-page PDF about 1.3¢.
Where is Nemotron 3 Ultra cheapest?
Among 5 providers we track, the lowest list price today is at DeepInfra: $0.40 input and $1.76 output per 1M tokens.
Does Nemotron 3 Ultra charge for reasoning tokens?
Yes. Reasoning (thinking) tokens are billed as output tokens. That is why the cost per task can be much higher than the visible answer suggests - our task costs include them.
Is the Nemotron 3 Ultra cost per task measured or estimated?
Estimated: Nemotron 3 Ultra was not run directly. Tokens come from Mistral Large 4 and are rescaled by the tokenizer of NVIDIA. Real usage can differ.