NVIDIA
Nemotron Cascade 2 30B A3B
最先端
Mar 2026公開日262K トークンコンテキストNVIDIA Open Model Licenseライセンス88 優秀品質
Nemotron Cascade 2 30B A3B (30B parameters) requires approximately 23.0 GB of VRAM with Q4_K_M quantization. As a Mixture of Experts model with 3B active parameters, it uses less memory than its total parameter count suggests. For the best balance of quality and speed, we recommend hardware with at least 27 GB of VRAM.
はじめに
— コピー&ペーストでローカル実行Copy-paste commands to run Nemotron Cascade 2 30B A3B on your machine.
Run
ollama run nemotron-cascade-2Quick specs
Parameters30B (3B active)
Architecturemoe (MoE)
Context262K tokens
Modalitytext
Min RAM11.7 GB
Rec. RAM18.3 GB (Q4_K_M)
LicenseNVIDIA Open Model License
FamilyNemotron
✓ Code✓ Chat✓ Reasoning
About this model
- •MoE: 30B total / 3B active — runs on single RTX 4090
- •Mamba-2 + Transformer hybrid architecture
- •Gold medal IMO 2025, IOI 2025, ICPC 2025
- •92.4% AIME 2025, 87.2% LiveCodeBench v6
- •262K context window
- •Thinking + instruct dual mode
関連モデル
おすすめ
最適なハードウェア
Nemotron Cascade 2 30B A3Bのおすすめ
このモデルを実行
量子化オプション
量子化レベル別VRAM推定値
No hardware detected — fit column shows raw VRAM estimates
| Quant | Bits | VRAM | Quality | Fit |
|---|---|---|---|---|
Q2_K | 2 | 11.7 GB | Low | — |
Q3_K_S | 3 | 14.7 GB | Low | — |
NVFP4 | 4 | 16.8 GB | Medium | — |
Q4_K_M | 4 | 18.3 GB | Medium | — |
Q5_K_M | 5 | 21.6 GB | High | — |
Q6_K | 6 | 24.6 GB | High | — |
Q8_0 | 8 | 32.1 GB | Very High | — |
F16 | 16 | 61.5 GB | Maximum | — |
Quality benchmarks
Nemotron Cascade 2 30B A3B benchmark scores
Coding
SWE-bench Verified—
HumanEval+—
Aider Polyglot—
LiveCodeBench87.2%
Reasoning
MMLU-Pro79.8%
GPQA Diamond76.1%
MATH-500—
ARC Challenge—
Source: official · 2026-03-19
ハードウェア互換性
全ハードウェアの適合度推定
Computing compatibility...
メモリ内訳
Reference: RTX 2060 6GB
Weights18.3 GB
KV Cache2.9 GB
Runtime1.2 GB
Headroom0.6 GB
よくある質問
FAQ — Nemotron Cascade 2 30B A3B
関連項目