FITMYLLM · NOVEMBER 4, 2026 · OPEN METHODOLOGY · COMMUNITY BENCHMARKS
LIVE · UPDATED DAILY
124.45B
Parameters (12B active)
Quantization Options Select your GPU for speed estimates Context length: 4K 8K 16K 32K 64K 128K 256K 512K 1M
Quant Bits VRAM @ 16K Quality IQ2_XXS 2.38 37.6 GB
37.5 + 0.1 KV
low IQ2_M 2.93 46.2 GB
46.1 + 0.1 KV
low Q2_K 3.16 49.7 GB
49.6 + 0.1 KV
low IQ3_XXS 3.25 51.1 GB
51.0 + 0.1 KV
low IQ3_XS 3.5 55.0 GB
54.9 + 0.1 KV
low Q3_K_S 3.64 57.2 GB
57.1 + 0.1 KV
low IQ3_M 3.76 59.1 GB
59.0 + 0.1 KV
low Q3_K_M 4 62.8 GB
62.7 + 0.1 KV
low Q3_K_L 4.3 67.5 GB
67.4 + 0.1 KV
moderate IQ4_XS 4.46 70.0 GB
69.9 + 0.1 KV
moderate Q4_K_S 4.67 73.2 GB
73.1 + 0.1 KV
moderate Q4_K_M 4.89 76.7 GB
76.6 + 0.1 KV
good Q5_K_S 5.57 87.2 GB
87.1 + 0.1 KV
good Q5_K_M 5.7 89.3 GB
89.2 + 0.1 KV
good Q6_K 6.56 102.6 GB
102.5 + 0.1 KV
excellent Q8_0 8.5 132.8 GB
132.7 + 0.1 KV
lossless FP16 16 249.5 GB
249.4 + 0.1 KV
lossless
Select your GPU above to see speed estimates and compatibility for each quantization.
Deploying for a team or in production? Size GPUs, cost & scaling in Enterprise → ▸ READY TO RUN THIS? RENT BY THE HOUR
RENT A GPU AND RUN NEMOTRON 3.5 SUPER VL 120B-A12B NOW
Spin up an A100 / H100 / 4090 in ~60s. Pay by the second. Cancel anytime.
Community Ratings Chat Coding Reasoning Creative Vision Roleplay Agentic
Loading ratings...
Run this model IQ2_XXS — 37.5 GB VRAM IQ2_M — 46.1 GB VRAM Q2_K — 49.6 GB VRAM IQ3_XXS — 51.0 GB VRAM IQ3_XS — 54.9 GB VRAM Q3_K_S — 57.1 GB VRAM IQ3_M — 59.0 GB VRAM Q3_K_M — 62.7 GB VRAM Q3_K_L — 67.4 GB VRAM IQ4_XS — 69.9 GB VRAM Q4_K_S — 73.1 GB VRAM Q4_K_M — 76.6 GB VRAM Q5_K_S — 87.1 GB VRAM Q5_K_M — 89.2 GB VRAM Q6_K — 102.5 GB VRAM Q8_0 — 132.7 GB VRAM FP16 — 249.4 GB VRAM
Ollama llama.cpp vLLM LM Studio KoboldCpp Jan Docker
▸ Easiest way to get started · Beginners
DOCS ↗ curl -fsSL https://ollama.com/install.sh | shCOPY
$ ollama run nemotron-3.5-super:120b-a12b-q4_K_MCOPY
Downloads and runs automatically. Add --verbose for speed stats.
▸ SETUP GUIDE >_
Auto-setup with fitmyllm CLI Detects your GPU, recommends the best model, downloads it, and starts chatting — zero config. Benchmarks your speed and contributes anonymous data to improve predictions.
Auto-detect GPU Live tok/s in chat Speed benchmarks 9 inference engines
GPUs that can run this model At Q4_K_M quantization. Sorted by minimum VRAM.
Find the best GPU for Nemotron 3.5 Super VL 120B-A12B
Build Hardware for Nemotron 3.5 Super VL 120B-A12B ▸ COLOPHON FITMYLLM · INDEPENDENT · DATA-DRIVEN
FITMYLLM · EST. 2025 · © 2026
RECOMMENDATIONS FROM PUBLISHED MATH, CORRECTED BY THE COMMUNITY — 30.
▸ SPEC SHEET
Nemotron 3.5 Super VL 120B-A12B — 124.45B MoE. ▸ SPECIFICATIONS
PARAMETERS 124.45B (12B active)
ARCHITECTURE Mixture of Experts
CONTEXT LENGTH 1024K tokens
CAPABILITIES chat, coding, reasoning, multilingual, vision, math, agentic, tool_use
RELEASE DATE 2026-10-15
PROVIDER NVIDIA
FAMILY nemotron ▸ VRAM REQUIREMENTS
QUANT BPW VRAM QUALITY IQ2_XXS 2.38 37.5 GB 65% IQ2_M 2.93 46.1 GB 75% Q2_K 3.16 49.6 GB 78% IQ3_XXS 3.25 51.0 GB 82% IQ3_XS 3.5 54.9 GB 84% Q3_K_S 3.64 57.1 GB 85% IQ3_M 3.76 59.0 GB 86% Q3_K_M 4 62.7 GB 88% Q3_K_L 4.3 67.4 GB 90% IQ4_XS 4.46 69.9 GB 92% Q4_K_S 4.67 73.1 GB 93% Q4_K_M 4.89 76.6 GB 94% Q5_K_S 5.57 87.1 GB 96% Q5_K_M 5.7 89.2 GB 96% Q6_K 6.56 102.5 GB 97% Q8_0 8.5 132.7 GB 100% FP16 16 249.4 GB 100%
§ 01 BENCHMARK SCORES
GPQA Diamond 89.5
LiveCodeBench 88.2
HLE 30.7
aa_scicode 47.8
aa_lcr 69.0
SWE-bench 71.1
aa_terminal_bench_v4 0.0
§ 02 RUN COMMAND
Run Nemotron 3.5 Super VL 120B-A12B locally with Ollama — needs 76.6 GB VRAM at Q4_K_M:
$ ollama run nemotron-3.5-super:120b
§ 03 COMPATIBLE GPUs
30 @ Q4_K_M Feedback 