BenchLeader

Nemotron 3 Nano 30B A3B

Reasoning effort
Quality#192 of 333
47.5Index
Below the median
−2.5 on the median
Price#10 of 249
$0.088per 1M tokens
Among the cheapest 10%
0.1× the median
Speed#20 of 215
186tokens/s
Top 10%
2.4× the median

1 cheaper ranked model scores higher: Qwen3.5-4B (49.4, $0.060).

As of 8 Nov 2026, Nemotron 3 Nano 30B A3B by NVIDIA scores 47.5 on the BenchLeader Index (593 ranked configurations), with its best results in instruction following (70) and weakest in reasoning (34). API pricing: $0.050 per 1M input tokens and $0.200 per 1M output tokens ($0.088 blended at 3:1), list price on 8 Nov 2026. It streams about 190 tokens/s (Artificial Analysis). Released 15 Dec 2025.

Key facts (as of 8 Nov 2026)

  • Nemotron 3 Nano 30B A3B API pricing: $0.050 per 1M input tokens and $0.200 per 1M output tokens ($0.088 blended at 3:1) (list price, 8 Nov 2026).
  • Nemotron 3 Nano 30B A3B streams about 190 tokens/s (Artificial Analysis, 8 Nov 2026).
  • Nemotron 3 Nano 30B A3B was released on 15 Dec 2025 (NVIDIA).

In detail. Nemotron 3 Nano 30B A3B is a NVIDIA open-weights reasoning model, released 15 Dec 2025. Its best configuration (thinking reasoning effort) ranks #397 of 593 on the BenchLeader Index at 47.5 ±11.6, in the lower half. The ± is the point: 461 other configurations score within that range, so they and this one cannot be told apart on quality alone — price and speed are what separate them. It scores highest in instruction following (70) and lowest in reasoning (34). At $0.088 per million tokens blended it is among the cheapest fifth of ranked models. Output speed of 186 tokens per second puts it in the fastest quarter, with a first answer in 8.0 s. It has been measured at 3 reasoning-effort settings; this summary describes the best-scoring one, and the tabs above switch between them. Last measured 15 Dec 2025.stale: no new result in six months

Blended price
$0.088/M
$0.050 in · $0.200 out
via Artificial Analysis
Output speed
186 tok/s
measured by Artificial Analysis
First answer
7.97 s
first token 0.46 s
Context
262k
via models.dev
Full answer
20 s
median, reasoning included
Cost per task
–
Tokens per task
–
Released
15 Dec 2025
How it scores by categoryDashed line = average model (50). One step of 15 = one standard deviation.
  1. Overall index48
  2. Reasoning34
  3. Agents & tools37
  4. Instruction following70
  5. Long context43

Get an email when NVIDIA ships a new model

And once when it gets ranked, plus the weekly review on Mondays.

Choose providers and emails

Reasoning-effort configurations

The same model behaves differently depending on how much it is allowed to think. Each row is one setting, scored only on the benchmarks that were run at that setting. “Not stated” collects results from publishers that did not say which setting they used; for a reasoning model that is usually its thinking mode, but we do not assume it. Pick a setting here or at the top of the page to see its price, speed and category scores.

EffortIndexRankSpeedFirst answerChat reply costCategories
no reasoning––143 tok/s0.95 s$0.0001Agents & tools 44 · Instruction following 41 · Long context 29
thinkingbest47.5#397186 tok/s7.97 s$0.0001Agents & tools 37 · Instruction following 70 · Long context 43 · Reasoning 34
not stated––139 tok/s0.46 s$0.0001Coding 48 · Human preference 47 · Maths 52 · Reasoning 47

How its index has moved

19 Sep 2026 to 28 Oct 2026
364860

The index is recomputed from scratch every day, so a line moves when a new benchmark result lands, when a publisher revises a score, or when the models it is normalised against change. Early movement usually means the score is still settling.

Benchmark results

See all LLM benchmarks. One column per reasoning effort. Rank is among every configuration of every model on that benchmark. Hover a score for the run it came from.

Metadata

Benchmarkno reasoningthinkingSourceTrend
AA Openness Indexnot in index83.3#1283.3#12Artificial Analysis

Reasoning

Benchmarkno reasoningthinkingnot statedSourceTrend
Arena Hard Promptseffort not stated––1326#247Arena (LMArena)
GPQA Diamond (AA)not in index39.9%#48975.7%#249–Artificial Analysis
Humanity's Last Exam (AA)not in index4.6%#50411.4%#305–Artificial Analysis
HLE-Diamond (AA)–3.5%#152–Artificial Analysis
CritPtnot in index0.0%#3290.9%#248–Artificial Analysis

Coding

Benchmarkno reasoningthinkingnot statedSourceTrend
Arena Codingeffort not stated––1361#239Arena (LMArena)
SciCode (AA)not in index–30.6%#222–Artificial Analysis
Terminal-Bench 4.0 (AA)not in index–0.0%#171–Artificial Analysis
Terminal-Bench 2.1 (AA)not in index–6.7%#203–Artificial Analysis
LiveCodeBench (AA)not in index36.0%#17874.1%#47–Artificial Analysis

Agents & tools

Benchmarkno reasoningthinkingSourceTrend
Terminal-Bench Hard12.1%#21513.6%#206Artificial Analysis
τ²-Bench Telecom (AA)not in index25.4%#29640.9%#225Artificial Analysis
GDPval-AA v2.1, normalised scorenot in index0.0%#1240.0%#124Artificial Analysis
GDPval-AA v2.1not in index-349#163262#144Artificial Analysis
τ³-Banking (AA)–6.0%#174Artificial Analysis
AutomationBenchnot in index–1.0%#215Artificial Analysis
GDP.pdf (AA)not in index–1.0%#225Artificial Analysis
AA-Briefcase v1.1not in index–78#161Artificial Analysis

Maths

Benchmarkno reasoningthinkingnot statedSourceTrend
Arena Mathseffort not stated––1348#205Arena (LMArena)
AIME 2025 (AA)not in index13.3%#6291.0%#3–Artificial Analysis

Knowledge

Benchmarkno reasoningSourceTrend
Global-MMLU-Lite (AA)not in index61.2%#32Artificial Analysis

Instruction following

Benchmarkno reasoningthinkingnot statedSourceTrend
IFBench37.5%#29771.1%#61–Artificial Analysis
Arena Instruction Followingnot in indexeffort not stated––1291#255Arena (LMArena)

Human preference

Benchmarknot statedSourceTrend
Arena Text1314#252Arena (LMArena)
Arena Creative Writingnot in index1242#279Arena (LMArena)
Arena Multi-turnnot in index1290#259Arena (LMArena)
Arena Longer Queriesnot in index1285#277Arena (LMArena)

Long context

Benchmarkno reasoningthinkingSourceTrend
AA-LCR v1.110.7%#48238.0%#381Artificial Analysis

Composite

Benchmarkno reasoningthinkingSourceTrend
AA Intelligence Index v4.3.2not in index6.8#5158.9#433Artificial Analysis
AA Progress Index v1not in index159.2#535214.4#367Artificial Analysis

API pricing: where to run it

Every provider serving this model through OpenRouter, with throughput and first-token latency measured on live traffic over the last 30 minutes and each provider’s own price. Purple marks the best in each column.

ProviderSpeedFirst tokenInput $/MOutput $/MContextQuantisation
Nebius Token Factory196 tok/s0.42 s$0.060$0.240262kfp8

What a task costs

Estimates from list price, output speed and time to first answer for the best configuration. “With caching” assumes three-quarters of the input is served from the prompt cache at $0.025 per 1M. Reasoning tokens are not modelled.

WorkloadTokens in / outCostWith cachingTime
Chat reply400 / 300$0.0001$0.00019.6 s
Summarize a 30-page report12,000 / 600$0.0007$0.000511.2 s
Code edit6,000 / 1,500$0.0006$0.000516.0 s
Agentic coding session60,000 / 4,000$0.0038$0.002729.5 s
Structured extraction2,000 / 200$0.0001$0.00019.0 s

More from NVIDIA

New this month

See also

Embed or cite

Show where Nemotron 3 Nano 30B A3B ranks on a model card or README. The badge redraws itself with every daily update, carries the data date and links back to this page. To see ranks like this on OpenRouter and Hugging Face model pages, get the Chrome extension.

[![Nemotron 3 Nano 30B A3B on BenchLeader: #188 overall](https://www.benchleader.com/badge/nemotron-3-nano-30b-a3b.svg)](https://www.benchleader.com/models/nemotron-3-nano-30b-a3b)

Cite this

The ranking changes daily, so a reference carries the data date. APA style:

BenchLeader. (2026, November 8). Nemotron 3 Nano 30B A3B on the BenchLeader Index (method v1.4) [Data set]. Retrieved November 8, 2026, from https://www.benchleader.com/models/nemotron-3-nano-30b-a3b

BibTeX:

@misc{benchleader_nemotron_3_nano_30b_a3b_20261108,
  author       = {{BenchLeader}},
  title        = {{Nemotron 3 Nano 30B A3B on the BenchLeader Index (method v1.4)}},
  year         = {2026},
  month        = nov,
  howpublished = {\url{https://www.benchleader.com/models/nemotron-3-nano-30b-a3b}},
  note         = {Data as of 2026-11-08}
}

Every badge style, the embeddable top-10 widget and how to cite the Index

Questions people ask

What are Nemotron 3 Nano 30B A3B's benchmark scores?
As of 8 Nov 2026, Nemotron 3 Nano 30B A3B (thinking effort) scores 47.5 ±11.6 on the BenchLeader Index, strongest in instruction following (70) and weakest in reasoning (34). That rests on independent benchmark results only, and every score on this page links to its publisher.
How does Nemotron 3 Nano 30B A3B rank?
Nemotron 3 Nano 30B A3B (thinking effort) is #397 of 593 ranked configurations on 8 Nov 2026. 461 other configurations score within its margin, so the exact position matters less than the band. The leader is Claude Opus 5.5 (max effort) at 69.4.
How much does Nemotron 3 Nano 30B A3B cost per 1M tokens?
As of 8 Nov 2026, Nemotron 3 Nano 30B A3B costs $0.050 per 1M input tokens and $0.200 per 1M output tokens, $0.088 blended (three input tokens per output token).
When was Nemotron 3 Nano 30B A3B released?
NVIDIA released Nemotron 3 Nano 30B A3B on 15 Dec 2025. BenchLeader has results for it at 3 reasoning-effort settings.

Data as of 8 Nov 2026. Compare these configurations.

Cite as: BenchLeader, “Nemotron 3 Nano 30B A3B: benchmarks, pricing, speed and rank”, data as of 8 Nov 2026, https://www.benchleader.com/models/nemotron-3-nano-30b-a3b.