whatmodel.sh / Nemotron-Labs-3 Puzzle 75B-A9B

Nemotron-Labs-3 Puzzle 75B-A9B

+ contribute

Nemotron-Labs-3 Puzzle 75B-A9B is an open-weights language model from NVIDIA, released June 2026. Its published size is 75.35B parameters (9.3B active per token). Licence: openmdw-1.1.

Catalogue identity notes

nvidia/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-BF16 @8fe5546 (created 2026-06-24, licence openmdw-1.1). NVIDIA Puzzle-compressed Nemotron 3 Super (June 2026): 75.3B total / 9.3B active (card). Lab rows from its own card table next to the catalogued Nemotron 3 Super 120B-A12B column (rule b, pooled in first-party-nvidia-scores).

Model information sources

Release date ↗ · Weights licence ↗

Card: 'reduces the model from 120.7B total / 12.8B active parameters to 75.3B total / 9.3B active parameters' (HF safetensors metadata 75,353,324,544).

64.2±1.2#155 of 719 LLMsfrom 12 benchmark results across 1 source

$ how it compares

The LLMs ranked either side of Nemotron-Labs-3 Puzzle 75B-A9B, out of 828 with a score, then more from NVIDIA.

#	model	score	where	RAM	released
175	Command A+ (05-2026) Cohere	64±1	api price unknown +local	~71GB	2026-05
176	Claude Haiku 4.5 Anthropic	64±1	api $2.00/M		2025-10
177	Nemotron-Labs-3 Puzzle 75B-A9B NVIDIA	64±1	local	~35GB	2026-06
178	Qwen3-VL 235B-A22B Thinking Alibaba	64±1	local	~111GB	2025-09
179	Qwen3-Coder 480B-A35B Instruct Alibaba	64±1	api $0.67/M +local	~78GB	2025-07
	more from NVIDIA
98	Nemotron 3 Ultra 550B-A55B NVIDIA	69±1	api $1.35/M +local	~89GB	2026-06
155	Nemotron Cascade 2 30B-A3B NVIDIA	66±1	local	~20GB	2026-03
168	Nemotron 3 Super 120B-A12B NVIDIA	65±1	api $0.28/M +local	~73GB	2026-03

see all LLMs · find the ones that fit your machine

$ what it's good at

agent tasks▁▁▃▆█▆▅▅▅▄▃▁ #135 of 300 · 51% of its score · 5 tests

Beats DeepSeek V3.2 Exp, Qwen3 235B-A22B Thinking 2507, o4-mini here, though they rank higher overall. Trails SmolLM3 3B, Jamba 1.5 Large, GPT-4o (2024-05-13) here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
τ²-bench Telecom lab	61.5% ✎	▁·▅█▆▅▂▃▃▂▃▁ beats 50/89	hard · avg 19%	▇▇▇▇▇▇▇▇▇ 22%	sharp · at its level
Terminal-Bench Hard lab	24.0% ✎	▂▅▆▂▂▆▂▂█▅▅▂ beats 18/27	very hard · avg 4.9%	▇▇▇▇▇ 12%	sharp · small run
SWE-bench Verified lab	56.9% ✎	▁▁▁▁▁▁▂▄█▆▁▁ beats 46/131	hard · avg 21%	▇▇▇ 8.4%	sharp · at its level · small run
τ²-bench Retail lab	63.2% ✎	▁····▂▁▃▆█▃▁ beats 15/44	medium · avg 35%	▇▇ 5.1%	at its level · small run
τ²-bench Airline lab	55.8% ✎	▁▁·▂▁▃▂▁█▅▂▃ beats 16/42	hard · avg 29%	▇▇ 4.1%	at its level · small run

coding▁▁▁▃▆██▇▆▄▃▁ #131 of 458 * · 12% of its score · 3 tests

Beats Qwen3.5 Plus, Gemini 2.5 Pro, Gemini 3.1 Flash-Lite here, though they rank higher overall. Trails Devstral Small 2 24B, Llama 3.1 Nemotron Ultra 253B v1, Nemotron 3.5 Lightning 30B-A3B here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
SWE-bench Verified lab	56.9% ✎	▁▁▁▁▁▁▂▄█▆▁▁ beats 46/131	hard · avg 21%	▇▇▇ 8.4%	sharp · at its level · small run
SciCode subproblems lab	40.6% ✎	▃··▃▃▃·▃·█·▅ beats 2/9	hard · avg 19%	▇ 3.3%	at its level · small run
LiveCodeBench v5 (Jul–Dec 2024) lab	81.1% ✎	█··········█ beats 0/1		—	not counted yet: no other model in the same run to compare it with

math▁▂▃▅▆▅▄█▄▁▁▁ #110 of 398 * · 17% of its score · 2 tests

Beats Claude Sonnet 4.6, MiniMax-M3, MiMo-V2.5 (310B-A15B) here, though they rank higher overall. Trails Gemini 2.5 Flash, Qwen3-VL 32B Thinking, gpt-oss-120b here, though they rank lower overall.

test	result	vs the others in its run	difficulty	counts	why
AIME 2025 lab	89.7% ✎	▁·▁▁▁▂▅▆█▇▅▁ beats 114/133	medium · avg 60%	▇▇▇▇ 10%	sharp
HMMT February 2025 lab	93.4% ✎	▁▁▁▃▁▅▅▃██▁▄ beats 22/31	easy · avg 72%	▇▇▇ 7.0%	small run

other tests4 not in a task · 28% of its score

test	result	vs the others in its run	difficulty	counts	why
GPQA (subset unspecified) lab	78.6% ✎	▂▁▂▂▄▃▆▃▅█▄▃ beats 92/104	hard · avg 48%	▇▇▇▇▇ 14%	sharp · at its level
Humanity's Last Exam lab	16.5% ✎	▁▁▁▃▆▆▇██▆▁▃ beats 38/104	very hard · avg 5.0%	▇▇ 6.1%	small run
MMLU-Pro lab	82.4% ✎	▁▁·▁▁▁▃▂▄▅█▁ beats 146/188	medium · avg 58%	▇▇ 5.6%	small run
IFBench prompt-level lab	71.9% ✎	█▄▄··▄·▄···▄ beats 1/6	medium · avg 55%	▇ 2.7%	small run

$ where to run it

llama.cpp · recommended

version

Install llama.cpp · Homebrew and winget installs name the commands llama-cli and llama-server instead.
macOS / Linux
curl -LsSf https://llama.app/install.sh | sh
Windows
irm https://llama.app/install.ps1 | iex
other ways to install ↗
Run Nemotron-Labs-3 Puzzle 75B-A9B · needs ~51.6 GB · from huggingface.co/RemySkye
chat in the terminal
llama cli -hf RemySkye/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-GGUF:Q4_K_M
Or start a local server and open the address it prints (http://localhost:9931)
llama serve -hf RemySkye/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-GGUF:Q4_K_M

LM Studio

version

Install LM Studio · Open the app once before using the lms command.
download it (macOS / Windows / Linux) ↗
Run Nemotron-Labs-3 Puzzle 75B-A9B · needs ~51.6 GB · from huggingface.co/RemySkye
open in LM Studio ↗
Or download it in the terminal
lms get https://huggingface.co/RemySkye/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-GGUF@Q4_K_M -y
Then run lms ls to see its name, and lms chat followed by that name

Ollama

Install Ollama
download it (macOS / Windows) ↗
Linux
curl -fsSL https://ollama.com/install.sh | sh
Run Nemotron-Labs-3 Puzzle 75B-A9B · needs ~31.5 GB · from huggingface.co/RemySkye
chat in the terminal
ollama run hf.co/RemySkye/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-GGUF:Q2_K

Other ways to run it

Run it as published · needs ~150.7 GB, NVIDIA only · from huggingface.co/nvidia (official)
vllm serve nvidia/NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-BF16 --tensor-parallel-size 2

Official weights and documentation: huggingface.co/nvidia ↗

$ how the score was worked out · NVIDIA vendor-reported benchm… 99%

Each result implies a score on its own, after adjusting for how hard its test is and who ran it. The 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's added up by who ran the tests. How it works

source	results	share of the score	licence
NVIDIA vendor-reported benchmark scores for its own Nemotron models (Hugging Face model cards) · lab-reported	12	▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇▇ 99%	Factual vendor-reported score (not licensed content)
prior · pull toward the middle		▇ 1%
weighted average		= 64.2

not counted (1)

NVIDIA vendor-reported benchmark scores for its own Nemotron models (Hugging Face model cards)

LiveCodeBench v5 (Jul–Dec 2024) [correct]

81.1%

fewer than three models have this test yet, or this source ran it for this model alone

$ improve this page

Update model information Add a place to run it Add a test result Propose a benchmark Correct a benchmark score Edit data on GitHub

Use a short GitHub form or edit the data through a pull request. Every number needs a source; here's which ones we can use.

data updated 2026-10-11 02:20 UTC · sources and licences · json