Nemotron-Cascade 2 30B-A3B — VRAM, speed & local setup
Written by Jakub Rusinowski · Last updated
31.6B parameters resident — about 19.9 GB at Q4_K_M, so a 24 GB card holds it — with 3.5B active per token, which is why it decodes far faster than its size suggests. 52 layers: 23 Mamba-2 state-space, 23 MoE, and 6 GQA attention. Each MoE layer carries 128 routed experts plus one shared, six routed activated per token. Gold-medal level at IMO 2025, IOI and the ICPC World Finals, reached through post-training rather than scale — and NVIDIA published that recipe. 1M-token context, NVIDIA Open Model License.
Nemotron-Cascade 2 30B-A3B needs about 20 GB of VRAM at Q4_K_M — quantized weights plus framework overhead, before any KV cache. On Apple Silicon that figure comes out of unified memory.
VRAM and speed by quantization
Quoted against NVIDIA RTX 4090 (24 GB). Weights plus overhead only: this model's architecture is unpublished, so no KV cache is added.
| Quant | Memory | VRAM | Speed (est.) | Fit |
|---|---|---|---|---|
| Q2_K 2.63 bpw | 11.2 GB | ~201 tok/s | Fits | |
| Q3_K_M 3.41 bpw | 14.3 GB | ~185 tok/s | Fits | |
| Q4_K_M 4.83 bpw | 19.9 GB | ~160 tok/s | Fits | |
| Q5_K_M 5.67 bpw | 23.2 GB | ~149 tok/s | Tight | |
| Q6_K 6.56 bpw | 26.7 GB | ~21 tok/s | Offload | |
| Q8_0 8.50 bpw | 34.4 GB | ~18 tok/s | Offload | |
| F16 16.00 bpw | 64 GB | — | Too big |
Black marker = usable memory on the NVIDIA RTX 4090 (24 GB). Estimates from the memory-bandwidth roofline on the methodology page. Nemotron-Cascade 2 30B-A3B VRAM calculator →
Get Nemotron-Cascade 2 30B-A3B running
The cheapest catalogued GPU that runs Nemotron-Cascade 2 30B-A3B is the AMD Radeon RX 7900 XTX (24 GB).
or compare on Vast.ai from $0.35/hr (typical low · varies)
As an Amazon Associate we earn from qualifying purchases. Cloud GPU links are referral links — we may earn a commission at no extra cost to you.
How to run Nemotron-Cascade 2 30B-A3B
Install Ollama, then run:
ollama run nemotron-cascade-2:30bSpecifications
Verified — Checked against the primary source — the model card or the vendor spec page — and corroborated by a second independent source.
- Parameters
- 31.6 Billion (3.5B active)
- Context window
- 1,048,576
- Architecture
- Hybrid Mamba-2 + MoE + GQA attention
- Provider
- NVIDIA
- Licence
- NVIDIA Open Model License
- Practical deployment
- Single consumer GPU
- Specified at
- Q4_K_M
- System RAM
- 32 GB
- Record updated
- 2026-09-11
Weights are downloadable and commercial use is permitted, subject to the licence’s acceptable-use terms.
Quality and use cases
Scores as published by the model’s authors or an independent evaluator — quality, not throughput, and not measured by us.
Can I run Nemotron-Cascade 2 30B-A3B on my GPU?
- Nemotron-Cascade 2 30B-A3B on AMD Radeon RX 7800 XT
- Nemotron-Cascade 2 30B-A3B on AMD Radeon RX 7900 XT
- Nemotron-Cascade 2 30B-A3B on AMD Radeon RX 7900 XTX
- Nemotron-Cascade 2 30B-A3B on AMD Radeon RX 9060 XT 16GB
- Nemotron-Cascade 2 30B-A3B on AMD Radeon RX 9060 XT 8GB
- Nemotron-Cascade 2 30B-A3B on AMD Radeon RX 9070
- Nemotron-Cascade 2 30B-A3B on AMD Radeon RX 9070 XT
- Nemotron-Cascade 2 30B-A3B on AMD Ryzen AI Max+ 395
- Nemotron-Cascade 2 30B-A3B on Apple M1
- Nemotron-Cascade 2 30B-A3B on Apple M1 Max
- Nemotron-Cascade 2 30B-A3B on Apple M1 Pro
- Nemotron-Cascade 2 30B-A3B on Apple M2
- Nemotron-Cascade 2 30B-A3B on Apple M2 Max
- Nemotron-Cascade 2 30B-A3B on Apple M2 Pro
- Nemotron-Cascade 2 30B-A3B on Apple M3
- Nemotron-Cascade 2 30B-A3B on Apple M3 Max
- Nemotron-Cascade 2 30B-A3B on Apple M3 Pro
- Nemotron-Cascade 2 30B-A3B on Apple M4
- Nemotron-Cascade 2 30B-A3B on Apple M4 Pro
- Nemotron-Cascade 2 30B-A3B on Apple M5
- Nemotron-Cascade 2 30B-A3B on Apple M5 Pro
- Nemotron-Cascade 2 30B-A3B on Intel Arc B570
- Nemotron-Cascade 2 30B-A3B on Intel Arc B580
- Nemotron-Cascade 2 30B-A3B on NVIDIA A100 80GB (PCIe)
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 3060 (12GB)
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 3070
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 3070 Ti
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 3080 (10GB)
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 3090
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 4060
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 4060 Ti 16GB
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 4070
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 4070 Super
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 4070 Ti
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 4070 Ti Super
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 4080
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 4080 Super
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 4090
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 5060
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 5060 Ti 16GB
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 5060 Ti 8GB
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 5070
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 5070 Ti
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 5080
- Nemotron-Cascade 2 30B-A3B on NVIDIA GeForce RTX 5090
- Nemotron-Cascade 2 30B-A3B on NVIDIA H100 80GB (PCIe)
- Nemotron-Cascade 2 30B-A3B on NVIDIA L40S
- Nemotron-Cascade 2 30B-A3B on NVIDIA RTX 6000 Ada Generation
- Nemotron-Cascade 2 30B-A3B on NVIDIA RTX PRO 6000 Blackwell
Nemotron-Cascade 2 30B-A3B — frequently asked questions
How much VRAM does Nemotron-Cascade 2 30B-A3B need?
About 20 GB at Q4_K_M — quantized weights plus framework overhead, before any KV cache. The cache grows with context length and is added on top; the table above folds it in. Apple Silicon counts unified memory toward the same figure.
Does Nemotron-Cascade 2 30B-A3B run on an RTX 4090 (24 GB)?
Yes. Nemotron-Cascade 2 30B-A3B needs about 20 GB at Q4_K_M, inside a 24 GB card, at an estimated 160 tokens/sec.
How do I run Nemotron-Cascade 2 30B-A3B locally?
Install Ollama and run `ollama run nemotron-cascade-2:30b`. That pulls the weights and starts a local OpenAI-compatible endpoint; after the download nothing leaves the machine.
What other sizes does Nemotron Cascade 2 come in?
Nemotron-Cascade 2 30B-A3B (20 GB), Nemotron Cascade 2 70B (43 GB). Every size shares the family's training and licence; the larger ones score higher and need proportionally more memory.