Nemotron 3 Nano 4B

NVIDIA · nvidia/nvidia-nemotron-3-nano-4b-bf16

ProviderNVIDIA
Familynemotron
Typellm-chat
Statusactive
Released2026-03-07
Updated2026-03-20
Parameters4B
Open weightsyes

Lineage

RelationshipModel
Is a finetune ofnvidia/NVIDIA-Nemotron-Nano-9B-v2

What it fits on

Ordered by predicted speed. Bandwidth sets decode rate; memory decides whether it runs at all.

Every figure here is computed, not measured. Fit is weights at each quantisation against device memory, with a 25% allowance for the KV cache, activations and the OS. Decode rate is the memory-bandwidth roofline at 70% efficiency. Nobody has run this model on these devices.
DeviceMemoryBandwidthBest qualityWeightstok/sFastest
Cerebras WSE-344.0 GB21000000.0 GB/sbf167.95 GB~1849728.2~7398912.7 q4
NVIDIA GB200 Grace Blackwell Superchip372.0 GB16000.0 GB/sbf167.95 GB~1409.3~5637.3 q4
AMD Instinct MI355X288.0 GB8000.0 GB/sbf167.95 GB~704.7~2818.6 q4
NVIDIA B200180.0 GB7700.0 GB/sbf167.95 GB~678.2~2712.9 q4
Google TPU7x (Ironwood)192.0 GB7380.0 GB/sbf167.95 GB~650.0~2600.2 q4
AMD Instinct MI325X256.0 GB6000.0 GB/sbf167.95 GB~528.5~2114.0 q4
AMD Instinct MI300X192.0 GB5300.0 GB/sbf167.95 GB~466.8~1867.3 q4
NVIDIA H200 SXM141.0 GB4800.0 GB/sbf167.95 GB~422.8~1691.2 q4
NVIDIA H100 NVL94.0 GB3900.0 GB/sbf167.95 GB~343.5~1374.1 q4
NVIDIA H100 SXM80.0 GB3350.0 GB/sbf167.95 GB~295.1~1180.3 q4
AMD Instinct MI250X128.0 GB3200.0 GB/sbf167.95 GB~281.9~1127.5 q4
Google TPU v5p95.0 GB2765.0 GB/sbf167.95 GB~243.5~974.2 q4
NVIDIA A100 80GB SXM80.0 GB2039.0 GB/sbf167.95 GB~179.6~718.4 q4
NVIDIA H100 PCIe80.0 GB2000.0 GB/sbf167.95 GB~176.2~704.7 q4
NVIDIA GeForce RTX 509032.0 GB1792.0 GB/sbf167.95 GB~157.8~631.4 q4
Google TPU v6e (Trillium)32.0 GB1638.0 GB/sbf167.95 GB~144.3~577.1 q4
AMD Instinct MI21064.0 GB1600.0 GB/sbf167.95 GB~140.9~563.7 q4
NVIDIA A100 40GB SXM40.0 GB1555.0 GB/sbf167.95 GB~137.0~547.9 q4
Google TPU v432.0 GB1200.0 GB/sbf167.95 GB~105.7~422.8 q4
NVIDIA GeForce RTX 409024.0 GB1008.0 GB/sbf167.95 GB~88.8~355.1 q4
AMD Radeon RX 7900 XTX24.0 GB960.0 GB/sbf167.95 GB~84.6~338.2 q4
NVIDIA GeForce RTX 508016.0 GB960.0 GB/sbf167.95 GB~84.6~338.2 q4
NVIDIA GeForce RTX 309024.0 GB936.0 GB/sbf167.95 GB~82.4~329.8 q4
NVIDIA L40S48.0 GB864.0 GB/sbf167.95 GB~76.1~304.4 q4
Apple M3 Ultra512.0 GB819.0 GB/sbf167.95 GB~72.1~288.6 q4
AMD Radeon RX 7900 XT20.0 GB800.0 GB/sbf167.95 GB~70.5~281.9 q4
Apple M2 Ultra192.0 GB800.0 GB/sbf167.95 GB~70.5~281.9 q4
Google TPU v5e16.0 GB800.0 GB/sbf167.95 GB~70.5~281.9 q4
NVIDIA GeForce RTX 4080 SUPER16.0 GB736.0 GB/sbf167.95 GB~64.8~259.3 q4
NVIDIA GeForce RTX 4070 Ti SUPER16.0 GB672.0 GB/sbf167.95 GB~59.2~236.8 q4
AMD Radeon RX 9070 XT16.0 GB640.0 GB/sbf167.95 GB~56.4~225.5 q4
Apple M4 Max128.0 GB546.0 GB/sbf167.95 GB~48.1~192.4 q4
Apple M1 Max64.0 GB400.0 GB/sbf167.95 GB~35.2~140.9 q4
Apple M2 Max96.0 GB400.0 GB/sbf167.95 GB~35.2~140.9 q4
Apple M3 Max128.0 GB400.0 GB/sbf167.95 GB~35.2~140.9 q4
NVIDIA GeForce RTX 3060 12GB12.0 GB360.0 GB/sbf167.95 GB~31.7~126.8 q4
NVIDIA L424.0 GB300.0 GB/sbf167.95 GB~26.4~105.7 q4
NVIDIA GeForce RTX 4060 Ti 16GB16.0 GB288.0 GB/sbf167.95 GB~25.4~101.5 q4
Apple M4 Pro64.0 GB273.0 GB/sbf167.95 GB~24.0~96.2 q4
Apple M432.0 GB120.0 GB/sbf167.95 GB~10.6~42.3 q4

Where it runs

Taken from the card's availability section.

PlatformId on that platform
Hugging Facenvidia/NVIDIA-Nemotron-3-Nano-4B-BF16

Verified benchmark evidence

Each row was checked against its source by a reviewer, and carries the model identifier as actually evaluated — which is not always the same as this card's.
BenchmarkEvaluated asScoreEvidence dateSource kind
aa_lcrNemotron 3 Nano 4B21.67%2026-09-10 evaluatedindependent evaluatorsource
critptNemotron 3 Nano 4B0.0%2026-09-10 evaluatedindependent evaluatorsource
gdpval_aaNemotron 3 Nano 4B0.0%2026-09-10 evaluatedindependent evaluatorsource
gpqa_diamondNemotron 3 Nano 4B51.31%2026-09-10 evaluatedindependent evaluatorsource

Reported benchmark scores

This card reports no benchmark scores yet.

BenchmarkCatalogue standingScore

Data

This card as JSON · See it in the graph · Edit on GitHub