[whatmodel.sh](../../README.md) / Nemotron-Labs-3 Puzzle 75B-A9B # Nemotron-Labs-3 Puzzle 75B-A9B NVIDIA · released 2026-06 · 75.35B params (9.3B active) · openmdw-1.1 **score 63.1** ±1.2 · #171 of 732 · from 12 benchmark results across 1 source · † no independently run results verified Score within the [LMArena (text)](https://whatmodel.sh/?set=lmarena-text&score=set) comparison set only. ## What it's good at ● marks Nemotron-Labs-3 Puzzle 75B-A9B among every model on that task or test, worst to best. 
task         vs the others          rank  counts  tests
agent tasks  ▁▁▃▆█▄●▅▄▄▂▁    #149 of 313     52%      5
coding       ▁▁▁▃▅▇█●▅▄▃▁   #139 of 469*     15%      3
math         ▁▁▃▅█▆▄●▄▂▁▁   #128 of 431*     16%      2
agent tasks · #149 of 313 · 52% of its score Beats [DeepSeek V3.2 Exp](deepseek-v3-2-exp.md), [o1 (2024-12-17)](o1-2024-12-17.md), [o4-mini](o4-mini.md) here, though they rank higher overall. Trails [SmolLM3 3B](smollm3-3b.md), [Jamba 1.5 Large](jamba-1-5-large.md), [GPT-4o (2024-05-13)](gpt-4o-2024-05-13.md) here, though they rank lower overall. 
test                         result  vs the others in its run   difficulty            counts  why
τ²-bench Telecom (maker)      61.5%  ▁·▅█▆●▂▃▃▂▃▁ beats 50/89   hard · avg 19%           21%  sharp · at its level
SWE-bench Verified (maker)    56.9%  ▁▁▁▁▁▁▂●█▆▁▁ beats 48/138  hard · avg 22%           12%  sharp · at its level · small run
Terminal-Bench Hard (maker)   24.0%  ▂▅▆▂▂▆▂▂●▅▅▂ beats 18/27   very hard · avg 5.2%     11%  sharp · small run
τ²-bench Retail (maker)       63.2%  ▁····▂▁▃●█▃▁ beats 15/44   medium · avg 35%        4.7%  at its level · small run
τ²-bench Airline (maker)      55.8%  ▁▁·▂▁▃▂▁●▅▂▃ beats 16/42   hard · avg 29%          3.8%  at its level · small run
coding · #139 of 469 · 15% of its score Beats [Qwen3.5 Plus](qwen3-5-plus.md), [Qwen3-Coder-Next (80B-A3B)](qwen3-coder-next.md), [Gemini 2.5 Pro](gemini-2-5-pro.md) here, though they rank higher overall. Trails [Llama 3.1 Nemotron Ultra 253B v1](llama-3-1-nemotron-ultra-253b.md), [Gemma 4 12B](gemma-4-12b.md), [Seed-OSS 36B Instruct](seed-oss-36b.md) here, though they rank lower overall. 
test                                result  vs the others in its run   difficulty      counts  why
SWE-bench Verified (maker)           56.9%  ▁▁▁▁▁▁▂●█▆▁▁ beats 48/138  hard · avg 22%     12%  sharp · at its level · small run
SciCode subproblems (maker)          40.6%  ▃··▃●▃·▃▃█·▅ beats 2/10    hard · avg 21%    3.8%  at its level · small run
LiveCodeBench v5 (Jul–Dec 2024) (…   81.1%  ●··········█ beats 0/1                          —  not counted yet: no other model in the …
math · #128 of 431 · 16% of its score Beats [Claude Sonnet 4.6](claude-sonnet-4-6.md), [MiniMax-M3](minimax-m3.md), [MiMo-V2.5 (310B-A15B)](mimo-v2-5.md) here, though they rank higher overall. Trails [Olmo 3.1 32B Think](olmo-3-1-32b-think.md), [Qwen3-VL 30B-A3B Thinking](qwen3-vl-30b-a3b-thinking.md), [Granite 4.2 8B](granite-4-2-8b.md) here, though they rank lower overall. 
test                        result  vs the others in its run    difficulty        counts  why
AIME 2025 (maker)            89.7%  ▁·▁▁▁▂▅▆█▇●▁ beats 114/133  medium · avg 61%     10%  sharp
HMMT February 2025 (maker)   93.4%  ▁▁▁▃▁▅▅▃█●▁▄ beats 22/31    easy · avg 75%      5.8%  small run
other tests (4) 
test                               result  vs the others in its run    difficulty            counts  why
GPQA (subset unspecified) (maker)   78.6%  ▂▁▂▂▄▃▆▃▅█●▃ beats 94/106   hard · avg 48%           13%  sharp · at its level
Humanity's Last Exam (maker)        16.5%  ▁▁▁▄▆▆●██▆▁▃ beats 41/111   very hard · avg 5.5%    7.0%  small run
MMLU-Pro (maker)                    82.4%  ▁▁·▁▁▁▃▂▄▅●▁ beats 146/188  medium · avg 59%        5.2%  small run
IFBench prompt-level (maker)        71.9%  ●▄▄··▄·▄···▄ beats 1/6      medium · avg 56%        2.3%  at its level · small run
## 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
NVIDIA vendor-reported benchmark scores…       12   99% ▇▇▇▇▇▇▇▇▇▇▇▇
prior: pull toward the middle                        1% ▇
weighted average                                   = 63.1
Not counted (1) 
source                  test                            score  why
NVIDIA vendor-reporte…  LiveCodeBench v5 (Jul–Dec 202…  81.1%  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-Labs-3 Puzzle 75B-A9B (Q4_K_M, needs ~51.6 GB, from [huggingface.co/RemySkye](https://huggingface.co/RemySkye/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-GGUF/tree/ff442763e62d5ef3cb8337c09826bb4a21924e34)) ```sh llama cli -hf RemySkye/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-GGUF:Q4_K_M ``` 3. Or start a local server and open the address it prints (http://localhost:9931) ```sh llama serve -hf RemySkye/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-GGUF:Q4_K_M ``` Other versions: Q8_0 (~83.5 GB) `llama cli -hf RemySkye/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-GGUF:Q8_0` · Q2_K (~31.5 GB) `llama cli -hf RemySkye/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-GGUF:Q2_K` ### 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-Labs-3 Puzzle 75B-A9B (Q4_K_M, needs ~51.6 GB, from [huggingface.co/RemySkye](https://huggingface.co/RemySkye/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-GGUF/tree/ff442763e62d5ef3cb8337c09826bb4a21924e34)) ```sh lms get https://huggingface.co/RemySkye/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-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 (~83.5 GB) `lms get https://huggingface.co/RemySkye/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-GGUF@Q8_0 -y` · Q2_K (~31.5 GB) `lms get https://huggingface.co/RemySkye/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-GGUF@Q2_K -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-Labs-3 Puzzle 75B-A9B (Q2_K, needs ~31.5 GB, from [huggingface.co/RemySkye](https://huggingface.co/RemySkye/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-GGUF/tree/ff442763e62d5ef3cb8337c09826bb4a21924e34)) ```sh ollama run hf.co/RemySkye/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-GGUF:Q2_K ``` ### Other ways to run it 1. Run it as published (BF16, needs ~150.7 GB, NVIDIA only, from [huggingface.co/nvidia](https://huggingface.co/nvidia/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-BF16/tree/8fe5546888e9bd03fdbf52d808adebdfca901b52)) ```sh vllm serve nvidia/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-BF16 --tensor-parallel-size 2 ``` Model repository and documentation: [huggingface.co/nvidia](https://huggingface.co/nvidia/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-BF16) [interactive version](https://whatmodel.sh/model/nemotron-labs-3-puzzle-75b-a9b) · [JSON](https://whatmodel.sh/api?format=json&all=1)