Model profile
Nemotron 3 Nano 30B A3B
Evidence summary
Safety. Nemotron 3 Nano 30B A3B has an estimated Safety rank of #191; its 90% source-sensitivity interval is #42–#316. Its behavior-only rank is #194; company governance moves the combined estimate to #191. Published Safety evidence spans 8 eval lineages and 4 of 7 components. Its strongest relative result is TAC (base_welfare_rate, #28 of 92); its weakest is Arena Factuality — Text Arena (factuality-only weighting) (factuality_bt_rating, #110 of 111).
Freedom. Nemotron 3 Nano 30B A3B has an estimated Freedom rank of #262; its 90% source-sensitivity interval is #138–#419. Published Freedom evidence spans 5 eval lineages and 1 of 1 components. Its strongest relative result is SM-Bench (adversarial, #9 of 92); its weakest is SM-Bench (overfit, #79 of 92).
Compare this model
Only models sharing at least one published sub-eval are listed.
Official and reference links
- Artificial Analysis ↗nvidia-nemotron-3-nano-30b-a3b
- Hugging Face ↗nvidia/NVIDIA-Nemotron-3-Nano-30B-A3B-BF16
- NVIDIA ↗qc69jvmznzxy/nemotron-3-nano-30b-a3b
- OpenRouter ↗nvidia/nemotron-3-nano-30b-a3b
- Official model catalog ↗Family-level model document · nvidia · first party
- Release source ↗curated outward evidence
Safety evals
Rank and direction are specific to the Safety portfolio. Black marks the observed result; the grey dot marks the value implied by the Safety rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Source |
|---|---|---|---|---|
| AA-Omnisciencehallucination_rate | #214 / 358 | ↓0.8327 | Source ↗official | |
| Arena Factuality — Text Arena (factuality-only weighting)factuality_bt_rating | #110 / 111 | ↑1378.0 | Source ↗official | |
| BullshitBench v2clear_pushback_rate | #85 / 122 | ↑0.19 | Source ↗official | |
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #62 / 104 | ↓49.1 | Source ↗official | |
| MT-JailBench CrescendoXsafety_score | #17 / 21 | ↑11.32 | Source ↗official | |
| SM-Benchadversarial | #83 / 92 | ↑73.66 | Source ↗official | |
| SM-Benchambiguous_interpretation | #82 / 92 | ↑71.43 | Source ↗official | |
| SM-Benchanti_hallucination | #87 / 92 | ↑75.13 | Source ↗official | |
| SM-Bencheq_boundaries | #78 / 92 | ↑53.93 | Source ↗official | |
| SM-Benchoverfit | #79 / 92 | ↑42.62 | Source ↗official | |
| TACbase_welfare_rate | #28 / 92 | ↑32.05 | Source ↗self run | |
| Vectara HHEM Factual Consistencyfactual_consistency_rate | #48 / 94 | ↑90.4 | Source ↗official |
Freedom evals
Rank and direction are specific to the Freedom portfolio. Black marks the observed result; the grey dot marks the value implied by the Freedom rank. Values stay on each source’s native scale.
| Eval / sub-eval | Rank | Value | Distribution | Source |
|---|---|---|---|---|
| Cisco AI Defense Rolling Single-Turn Leaderboardsingle_turn_attack_success_rate | #43 / 104 | ↑49.1 | Source ↗official | |
| MT-JailBench CrescendoXsafety_score | #3 / 21 | ↓11.32 | Source ↗official | |
| SM-Benchadversarial | #9 / 92 | ↓73.66 | Source ↗official | |
| SM-Bencheq_boundaries | #78 / 92 | ↑53.93 | Source ↗official | |
| SM-Benchoverfit | #79 / 92 | ↑42.62 | Source ↗official | |
| SpeechMap model completioncomplete_pct | #65 / 181 | ↑67.7 | Source ↗official | |
| UGI Leaderboard — base-model willingnesswillingness_adherence_score | #70 / 156 | ↑2.25 | Source ↗official | |
| UGI Leaderboard — base-model willingnesswillingness_direct_score | #66 / 156 | ↑3.5 | Source ↗official |
Values evaluations
Descriptive values and political-framing results are separate from safety/ethics ranks. Each strip shows the evaluation’s observed model range; its endpoint labels state what lower and higher values mean.