Models / Nemotron 3.5 Bolt 74B-A9B / VRAM
How much VRAM do you need for Nemotron 3.5 Bolt 74B-A9B?
Last updated:
Weight math from the model card, plus every measured RigBase result for Nemotron 3.5 Bolt 74B-A9B. Leave extra room for context (KV cache).
Weight size by quant
| Quant | Bits/weight | Weights only |
|---|---|---|
| Q8_0 | 8.50 | 78.7 GB |
| Q6_K | 6.56 | 60.7 GB |
| Q5_K_M | 5.70 | 52.8 GB |
| Q4_K_M | 4.89 | 45.3 GB |
| Q3_K_M | 4.00 | 37.0 GB |
| Q2_K | 3.16 | 29.2 GB |
Bits-per-weight for GGUF rows: llama.cpp quantize README. Parameter count: via llamabench.ai. We advise staying at Q4 or above for chat quality — see Is Q4 good enough?.
Measured speeds for Nemotron 3.5 Bolt 74B-A9B
| Hardware | Quant | Engine | tok/s |
|---|---|---|---|
| RTX 5090 · 32 GB | NVIDIA mixed 2-bit IQ2 GGUF (22.0 GB) | llama.cpp build 11669 | 154.7 |
| RTX 4090 · 24 GB | llama-bench tg128 (quant per result) | llama.cpp | 108 |
| RTX 4090 · 24 GB | NVIDIA mixed 2-bit IQ2 GGUF (22.0 GB) | llama.cpp build 11669 | 107.3 |
| RTX 3090 · 24 GB | NVIDIA mixed 2-bit IQ2 GGUF (22.0 GB) | llama.cpp build 11669 (620d2225) | 87.75 |
| Nvidia GB10 (unified) · 128 GB | IQ2_XXS + NVFP4 + LiLiCorr | TensorFold Zig, 1 stream | 71 ⚡ |
| Nvidia GB10 (unified) · 128 GB | 2-bit 24GB GGUF + DFlash speculative decoding (7 d | llama.cpp | 61.61 ⚡ |
| Nvidia GB10 (unified) · 128 GB | 2-bit 24GB GGUF, no speculative decoding | llama.cpp | 39.47 |
⚡ = speculative decoding (MTP, DFlash…). Full list: best home rig for Nemotron 3.5 Bolt 74B-A9B.
Related
All Nemotron 3.5 Bolt 74B-A9B builds VRAM + KV calculator By VRAM size GGUF vs AWQ vs FP8 vs NVFP4 How much system RAM?
FAQ
How much VRAM do you need for Nemotron 3.5 Bolt 74B-A9B?
Weights alone are about 45.3 GB at Q4_K_M (74.04B × 4.89 bits ÷ 8). Add memory for the KV cache (context) and the runtime. The smallest-memory measured result in RigBase is 24 GB (RTX 4090) at 108 tok/s. Going below Q4 usually hurts quality for chat; see the Q4 guide.
What is the fastest measured speed for Nemotron 3.5 Bolt 74B-A9B?
RTX 5090: 154.7 tok/s plain decode (NVIDIA mixed 2-bit IQ2 GGUF (22.0 GB), llama.cpp build 11669).
Does system RAM replace VRAM?
Only for overflow. When the whole model fits in GPU or unified memory, system RAM barely matters for decode speed. See the system-RAM guide.
Referral link: if you sign up with Google and add $10, Runpod gives you and RigBase bonus credit (terms). It never affects our data.