[whatmodel.sh](../../README.md) / Nemotron 3 Nano 4B # Nemotron 3 Nano 4B NVIDIA · released 2026-03 · 3.97B params · nvidia-open-model **score 54.1** ±0.8 · #301 of 732 · from 26 benchmark results across 2 sources Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only. ## What it's good at ● marks Nemotron 3 Nano 4B among every model on that task or test, worst to best. 
task            vs the others          rank  counts  tests
agent tasks     ▁▁▃●█▄▅▅▄▄▂▁    #255 of 313     22%      7
coding          ▁▁▁▃▅●█▇▅▄▃▁    #292 of 469     18%      6
chat & writing  ▁▁▁▃▆▄▆█●▇▇▃   #167 of 402*    0.3%      2
math            ▁▁▃▅█●▄█▄▂▁▁    #193 of 431     38%      4
web search      ●·▅▅█·▃▃▃·▅▃     #15 of 15*    2.1%      1
agent tasks · #255 of 313 · 22% of its score Beats [DeepSeek R1 0528 Qwen3 8B](deepseek-r1-0528-qwen3-8b.md), [Sky-T1 32B Preview](sky-t1-32b-preview.md), [Magistral Small 1.0 (2506)](magistral-small-2506.md) here, though they rank higher overall. Trails [Granite 4.0 1B](granite-4-0-1b.md), [SmolLM3 3B](smollm3-3b.md), [Qwen3-VL 2B Instruct](qwen3-vl-2b-instruct.md) here, though they rank lower overall. 
test                               result  vs the others in its run    difficulty            counts  why
τ²-bench Telecom (maker)            33.0%  ▁·▅●▆▅▂▃▃▂▃▁ beats 27/89    hard · avg 19%           10%  sharp · at its level
BFCL V4 (December 2025 archive)     43.7%  ▄·●·▄▄▄▄·▄·▄ beats 2/8      medium · avg 43%        2.9%  at its level
SWE-bench Verified                   3.0%  ▄··▄●█··▄·▄█ beats 2/8      very hard · avg 11%     2.4%  sharp · noisy
BrowseComp-ZH                        3.3%  ●▄·▄▄▄·█···█ beats 0/8      very hard · avg 13%     2.1%  noisy
SWE-bench Pro (split unspecified)    0.1%  ●▂···▂···▄·▂ beats 1/8      very hard · avg 3.9%    1.6%  sharp · too hard for it · noisy
τ³-bench Banking                     1.2%  ●▂·▅█▂▃▃▃▆▂▅ beats 0/24     very hard · avg 3.6%    1.2%  blunt · too hard for it
BFCL v3 (maker)                     61.1%  ▁▁▁▁▁▃▅█●▅▄▁ beats 102/171  medium · avg 58%        1.0%  blunt · at its level · small run
coding · #292 of 469 · 18% of its score Beats [Qwen3-VL 32B Instruct](qwen3-vl-32b-instruct.md), [Solar Open 100B (102B-A12B)](solar-open-100b.md), [Qwen3-VL 30B-A3B Instruct](qwen3-vl-30b-a3b-instruct.md) here, though they rank higher overall. Trails [DeepSeek LLM 67B Chat](deepseek-llm-67b-chat.md), [Yi 1.5 9B Chat](yi-1-5-9b-chat.md), [Mistral NeMo Instruct (2407)](mistral-nemo-instruct-2407.md) here, though they rank lower overall. 
test                                result  vs the others in its run  difficulty            counts  why
LiveCodeBench v6 (window unspecif…   50.7%  ▃····█·●▅▃·▃ beats 4/8    medium · avg 38%        5.5%  at its level
LiveCodeBench Pro 2025Q2 (Easy)      51.6%  ▄···█▄··▄●▄▄ beats 5/8    hard · avg 33%          3.9%  at its level
LiveCodeBench Pro 2025Q2 (Medium)     5.3%  █···▂··●▂··▂ beats 5/8    very hard · avg 2.0%    2.7%  sharp · too hard for it · noisy
SWE-bench Verified                    3.0%  ▄··▄●█··▄·▄█ beats 2/8    very hard · avg 11%     2.4%  sharp · noisy
LiveCodeBench (version unspecifie…   51.8%  ▁···▁▁▁█●▃▄▁ beats 18/38  medium · avg 46%        1.8%  at its level · small run
SWE-bench Pro (split unspecified)     0.1%  ●▂···▂···▄·▂ beats 1/8    very hard · avg 3.9%    1.6%  sharp · too hard for it · noisy
chat & writing · #167 of 402 · 0.3% of its score Beats [MiniMax-M2](minimax-m2.md), [MiniCPM5 2B](minicpm5-2b.md), [Granite 4.2 30B](granite-4-2-30b.md) here, though they rank higher overall. Trails [LFM2.5 1.2B Thinking](lfm2-5-1-2b-thinking.md), [LFM2.5 8B-A1B](lfm2-5-8b-a1b.md), [LFM2.5 2.6B](lfm2-5-2-6b.md) here, though they rank lower overall. 
test                                result  vs the others in its run  difficulty      counts  why
IFEval (aggregation unspecified)     88.0%  ▄▄····▄··●██ beats 4/8    easy · avg 76%    0.3%  noisy
IFEval prompt-level (strictness u…   87.9%  no other model yet                             —  not counted yet: no other model in the …
math · #193 of 431 · 38% of its score Beats [MiniMax-M2.1](minimax-m2-1.md), [DeepSeek V3 0324](deepseek-v3-0324.md), [DeepSeek V3 (Dec '24)](deepseek-v3.md) here, though they rank higher overall. Trails [Mixtral 8x7B v0.1 (base)](mixtral-8x7b-v0-1.md), [Phi-4-mini-reasoning 3.8B](phi-4-mini-reasoning.md), [DeepSeek R1 Distill Qwen 7B](deepseek-r1-distill-qwen-7b.md) here, though they rank lower overall. 
test                result  vs the others in its run  difficulty           counts  why
AIME 2025            56.3%  █▄▄▄●····█·▄ beats 5/8    medium · avg 61%        16%  sharp · at its level · several runs
HMMT February 2026   51.3%  ▅··▃▃·▃··●·█ beats 5/8    hard · avg 30%          10%  sharp · at its level
AIME 2026            62.1%  ▄·▄█·▄●···█▄ beats 5/8    medium · avg 48%       6.7%  sharp · at its level
MATH-500             91.6%  ██●▄▄··▄···▄ beats 4/8    very easy · avg 89%    5.6%  maxed out · several runs
web search · #15 of 15 · 2.1% of its score Trails [Gemma 4 E2B](gemma-4-e2b.md), [Qwen3.5 2B](qwen3-5-2b.md), [LFM2.5 8B-A1B](lfm2-5-8b-a1b.md) here, though they rank lower overall. 
test           result  vs the others in its run  difficulty           counts  why
BrowseComp-ZH    3.3%  ●▄·▄▄▄·█···█ beats 0/8    very hard · avg 13%    2.1%  noisy
other tests (8) 
test                                result  vs the others in its run   difficulty           counts  why
GPQA (subset unspecified) (maker)    53.2%  ▂▁▂▂▄▃▆●▅█▃▃ beats 55/106  hard · avg 48%         7.6%  sharp · at its level
MMLU-Pro                             65.7%  ▂··▂●·▂▂···▂ beats 4/8     medium · avg 59%       4.1%  at its level
MMLU-Redux                           79.8%  ▂···●··▂▂··▂ beats 2/8     easy · avg 76%         3.6%  typical
OJBench                              20.0%  ▃····▃▅··●▃▃ beats 5/8     very hard · avg 12%    2.8%  typical
SuperGPQA                            37.8%  ▃·▃·▃●█····▃ beats 3/8     hard · avg 32%         2.5%  at its level
LongBench v2                         32.0%  ▃······▅●▃·█ beats 3/8     very hard · avg 32%    2.3%  typical
IFBench (aggregation unspecified)    58.3%  ▅····▃▃·●·▃▃ beats 4/8     medium · avg 44%       1.7%  at its level
Multi-IF (aggregation unspecified)   65.9%  ▄▄···▄·●·▄██ beats 3/8     medium · avg 60%       1.2%  at its level
## How the score was worked out Each result implies a score on its own, after adjusting for how hard that test is and who ran it. The model's score is their weighted average, with a small pull toward the middle for thinly tested models. Each test's weight is its *counts* column above; here it is added up by who ran the tests. [Method](../method.md) 
source                                    results  share of the score
openbmb/MiniCPM5-2B model card                 20   68% ▇▇▇▇▇▇▇▇
NVIDIA vendor-reported benchmark scores…        6   32% ▇▇▇▇
prior: pull toward the middle                        0% ▇
weighted average                                   = 54.1
Licences: openbmb/MiniCPM5-2B model card ([Apache-2.0](https://www.apache.org/licenses/LICENSE-2.0))
Not counted (6) 
source                  test                            score  why
nvidia/NVIDIA-Nemotro…  AIME 2025                       78.5%  only model in this source's run
nvidia/NVIDIA-Nemotro…  MATH-500                        95.4%  only model in this source's run
nvidia/NVIDIA-Nemotro…  GPQA (subset unspecified)       53.2%  only model in this source's run
nvidia/NVIDIA-Nemotro…  LiveCodeBench (version unspec…  51.8%  only model in this source's run
nvidia/NVIDIA-Nemotro…  BFCL v3                         61.1%  only model in this source's run
nvidia/NVIDIA-Nemotro…  IFEval prompt-level (strictne…  87.9%  too few models have it yet
## How to run it ### llama.cpp (recommended) 1. Install llama.cpp. Homebrew and winget installs name the commands llama-cli and llama-server instead. [Other ways to install](https://github.com/ggml-org/llama.cpp/blob/master/docs/install.md) ```sh # macOS / Linux curl -LsSf https://llama.app/install.sh | sh # Windows irm https://llama.app/install.ps1 | iex ``` 2. Run Nemotron 3 Nano 4B (Q4_K_M, needs ~2.8 GB, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/NVIDIA-Nemotron-3-Nano-4B-GGUF/tree/f4732c3cd1c961777138b226030e6c17f6f7db84)) ```sh llama cli -hf lmstudio-community/NVIDIA-Nemotron-3-Nano-4B-GGUF:Q4_K_M ``` 3. Or start a local server and open the address it prints (http://localhost:9931) ```sh llama serve -hf lmstudio-community/NVIDIA-Nemotron-3-Nano-4B-GGUF:Q4_K_M ``` Other versions: Q8_0 (~4.2 GB) `llama cli -hf lmstudio-community/NVIDIA-Nemotron-3-Nano-4B-GGUF:Q8_0` ### LM Studio 1. Install LM Studio. Open the app once before using the lms command. [Download it (macOS / Windows / Linux)](https://lmstudio.ai/download) 2. Run Nemotron 3 Nano 4B (Q4_K_M, needs ~2.8 GB, from [huggingface.co/lmstudio-community](https://huggingface.co/lmstudio-community/NVIDIA-Nemotron-3-Nano-4B-GGUF/tree/f4732c3cd1c961777138b226030e6c17f6f7db84)) ```sh lms get https://huggingface.co/lmstudio-community/NVIDIA-Nemotron-3-Nano-4B-GGUF@Q4_K_M -y ``` 3. Then run lms ls to see its name, and lms chat followed by that name Other versions: Q8_0 (~4.2 GB) `lms get https://huggingface.co/lmstudio-community/NVIDIA-Nemotron-3-Nano-4B-GGUF@Q8_0 -y` · 4bit (~2.2 GB, Mac only) `lms get https://huggingface.co/mlx-community/NVIDIA-Nemotron-3-Nano-4B-4bit -y` · 8bit (~4.2 GB, Mac only) `lms get https://huggingface.co/mlx-community/NVIDIA-Nemotron-3-Nano-4B-8bit -y` ### Ollama 1. Install Ollama. [Download it (macOS / Windows)](https://ollama.com/download) ```sh # Linux curl -fsSL https://ollama.com/install.sh | sh ``` 2. Run Nemotron 3 Nano 4B (Q4_K_M, needs ~2.88 GB, from [ollama.com](https://ollama.com/library/nemotron-3-nano/tags)) ```sh ollama run nemotron-3-nano:4b ``` Other versions: Q8_0 (~5.26 GB) `ollama run nemotron-3-nano:4b-q8_0` ### MLX (Mac) 1. Install MLX (Mac). Apple silicon Macs only, text only. ```sh # with uv uv tool install mlx-lm # or pip install mlx-lm ``` 2. Run Nemotron 3 Nano 4B (4bit, needs ~2.2 GB, Mac only, from [huggingface.co/mlx-community](https://huggingface.co/mlx-community/NVIDIA-Nemotron-3-Nano-4B-4bit/tree/c4d79ba1901d99806ef757642a552acebb851a35)) ```sh mlx_lm.chat --model mlx-community/NVIDIA-Nemotron-3-Nano-4B-4bit ``` 3. Or start an OpenAI-compatible server at http://localhost:8080/v1 ```sh mlx_lm.server --model mlx-community/NVIDIA-Nemotron-3-Nano-4B-4bit ``` Other versions: 8bit (~4.2 GB, Mac only) `mlx_lm.chat --model mlx-community/NVIDIA-Nemotron-3-Nano-4B-8bit` Model repository and documentation: [huggingface.co/nvidia](https://huggingface.co/nvidia/NVIDIA-Nemotron-3-Nano-4B-BF16) ## Published speed measurements Reference results for the hardware, engine and settings below. These are not verified speeds of the downloads above or predictions for your machine. 
hardware                      precision         speed  engine and settings
NVIDIA DGX Spark (GB10)       Q8_0        52.85 tok/s  llama.cpp 1bbec6a75 (8382), CUDA …
NVIDIA DGX Spark (GB10)       Q8_0        52.32 tok/s  llama.cpp 1bbec6a75 (8382), CUDA …
NVIDIA DGX Spark (GB10)       Q8_0        51.27 tok/s  llama.cpp 1bbec6a75 (8382), CUDA …
NVIDIA DGX Spark (GB10)       Q8_0        49.71 tok/s  llama.cpp 1bbec6a75 (8382), CUDA …
NVIDIA DGX Spark (GB10)       Q8_0        46.71 tok/s  llama.cpp 1bbec6a75 (8382), CUDA …
Apple M5 Max (40-core GPU),…  q4_k_m     104.03 tok/s  ollama (version not recorded) · l…
Each speed links to its source, which has the full settings. [interactive version](https://whatmodel.sh/model/nemotron-3-nano-4b) · [JSON](https://whatmodel.sh/api?format=json&all=1)