NVIDIA Nemotron 3 Nano 4B
Mar 2026
NVIDIA Nemotron 3 Nano 4BCompare inon
Benchmarks of NVIDIA Nemotron 3 Nano 4B18
BenchmarkByScoreRank
Overall1
Knowledge4
Science3
Coding3
Agents3
Documents2
Legacy · saturated or no longer reported2
No benchmark matches
—per task?
self-host onlyprice
—output speed?
262Kcontext
Ranking?
Alternatives?
ModelScoreCostPer taskWhat changes
BetterGPT-6 Lunalow×4.2—$0.005Function calling only with none · Chat Completions+3IOpenK2 Horizon 7Bthinking×3.7—$0.30 / $1.20Memory (4-bit): 2 GB → 4.5 GBBetterGPT-5.6 Lunalow×4.1—$0.01BetterGPT OSS 20BhighOpen×2.3—$0.011Context: 262K → 131K+1
BetterMistral Small 4thinkingOpen×2—$0.015Memory (4-bit): 2 GB → 72 GBModelScoreCostPer taskWhat changes
Nothing here yet.
ModelScoreCostPer taskWhat changes
GPT-6 Lunalow×4.2—$0.005Function calling only with none · Chat Completions+3GPT-5.6 Lunalow×4.1—$0.01GPT OSS 20BhighOpen×2.3—$0.011Context: 262K → 131K+1Mistral Small 4thinkingOpen×2—$0.015Memory (4-bit): 2 GB → 72 GBNemotron 3 Nano 30B A3BthinkingOpen×2.3—$0.017Context: 262K → 131K+1ModelScoreCostPer taskWhat changes
IK2 Horizon 7BthinkingOpen×3.7—$0.30 / $1.20Memory (4-bit): 2 GB → 4.5 GB
Gemma 4 12BthinkingOpen×3.2—$0.10 / $0.30Context: 262K → 256K+1
Qwen 3.5 9BthinkingOpen×3.1—$0.21Memory (4-bit): 2 GB → 4.8 GBIK2 Horizon 3.7BthinkingOpen×2.8—$0.30 / $1.20Memory (4-bit): 2 GB → 2.5 GBTFalcon H1R 7BthinkingOpen×2.3——Context: 262K → 256K+1ModelScoreCostPer taskWhat changes
Nemotron 3.5 Super VLOpen×6.3—$0.19Memory (4-bit): 2 GB → 75 GBNemotron 3 UltrathinkingOpen×5.8—$0.58Memory (4-bit): 2 GB → 336 GBNemotron 3.5 Bolt 74B-A9BOpen×5.1——Context: 262K → 256K+1Nemotron 3 Super 120B A12BOpen×4.3—$1.64Memory (4-bit): 2 GB → 74 GBNemotron Cascade 2 30B A3BOpen×2.5——Memory (4-bit): 2 GB → 19 GBNemotron 3.5 LightningthinkingOpen×2.2—$0.079Memory (4-bit): 2 GB → 20 GBLlama 3.1 Nemotron Nano 4B v1.1Opensame——Context: 262K → 128K+1Nemotron 3 Nano Omni 30B A3B ReasoningOpensame—$0 / $0Context: 262K → 256K+1All NVIDIA models →Run it yourself
estimateFits on✓ 8 GB (phone, laptop) · full quality✓ 16 GB · full quality✓ 24 GB GPU · full quality✓ 48 GB GPU · full quality✓ 80 GB GPU · full quality
Licencenvidia-nemotron-open-model-licensecommercial use with conditions
Benchmarks
all categoriesChat & Knowledge
Humanity's Last Exam4.9#282
AA Omniscience Index-70.5#295
CritPt (physics)0#180
AA Intelligence Index 4.3.24.8#336
AA Omniscience · Accuracy9.6#283
AA-LCR (long context) 1.121.7#234
GDP.pdf · AA all-pass0.4#128
GPQA Diamond51.3#259
Coding
SciCode16.1#137
Terminal-Bench 2.13.7#140
Agents & Tool Use
AutomationBench-AA0.4#122
τ²-bench · AA28.1#154
Writing & Language
IFBench (instructions)58.2#89
FAQ
How much does NVIDIA Nemotron 3 Nano 4B cost?›
There is no per-unit API price: self-host only.
Is NVIDIA Nemotron 3 Nano 4B free to use?›
There is no free API. The weights are open, so you can run it yourself.
How good is NVIDIA Nemotron 3 Nano 4B, and what is it best at?›
#280 of 334 in Chat & Knowledge, #75 of 132 in Coding, #122 of 133 in Agents & Tool Use, #89 of 245 in Writing & Language, #128 of 133 in Vision & Documents.
Is NVIDIA Nemotron 3 Nano 4B better than Llama 3.1 Nemotron Nano 4B v1.1?›
On Humanity's Last Exam, NVIDIA Nemotron 3 Nano 4B scores 4.9 vs 5.2 — about the same.
What is the context window of NVIDIA Nemotron 3 Nano 4B?›
262K tokens.
When was NVIDIA Nemotron 3 Nano 4B released? Is there a newer version?›
Released 2026-03-16. It is the current version in its line.
Is NVIDIA Nemotron 3 Nano 4B open source?›
Yes, the weights are open (nvidia-nemotron-open-model-license licence).
Can I run NVIDIA Nemotron 3 Nano 4B locally?›
The weights are open and can be self-hosted; hardware needs are not confirmed yet.
Can NVIDIA Nemotron 3 Nano 4B read images?›
No, it accepts text only.
What is NVIDIA Nemotron 3 Nano 4B?›
NVIDIA Nemotron 3 Nano 4B is an open-weights language model from NVIDIA, released Mar 2026, the successor to Llama 3.1 Nemotron Nano 4B v1.1.
What are the best alternatives to NVIDIA Nemotron 3 Nano 4B?›
GPT-6 Luna (better), K2 Horizon 7B (open weights).