{
 "id": "nvidia-nemotron-3.5-lightning-30b-a3b",
 "name": "NVIDIA-Nemotron-3.5-Lightning 30B-A3B",
 "repo": "nvidia/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16",
 "lab": {
  "name": "NVIDIA",
  "country": "United States"
 },
 "released": "2026-08-01",
 "parameters": 31.577937344,
 "activeParameters": 3,
 "maxContext": 262144,
 "architecture": "52 layers · 52 full attention (2 KV × 128)",
 "page": "https://studiotvai.com/vram-requirements/nvidia-nemotron-3-5-lightning-30b-a3b",
 "calculator": "https://studiotvai.com/llm-vram-calculator?model=nvidia-nemotron-3.5-lightning-30b-a3b",
 "weightsGiB": {
  "bf16": 56.33,
  "fp8": 28.82,
  "q4_k_m": 21.16
 },
 "kvCacheGiBPerRequest": [
  {
   "context": 8192,
   "bf16": 0.406,
   "fp8": 0.203
  },
  {
   "context": 32768,
   "bf16": 1.625,
   "fp8": 0.813
  },
  {
   "context": 131072,
   "bf16": 6.5,
   "fp8": 3.25
  },
  {
   "context": 262144,
   "bf16": 13,
   "fp8": 6.5
  }
 ],
 "recurrentStateGiBPerRequest": 0,
 "gpusNeeded": {
  "scenario": "one request of 8192 tokens, BF16 KV cache; vLLM on data-center GPUs, llama.cpp elsewhere",
  "gpus": [
   {
    "gpu": "rtx-4090",
    "name": "GeForce RTX 4090",
    "memoryGB": 24,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 3,
     "fp8": 2,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "rtx-5090",
    "name": "GeForce RTX 5090",
    "memoryGB": 32,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 2,
     "fp8": 1,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "rtx-pro-6000",
    "name": "RTX PRO 6000 Blackwell",
    "memoryGB": 96,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 1,
     "fp8": 1,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "mac-m5-max-128",
    "name": "Mac M5 Max 128 GB",
    "memoryGB": 96,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 1,
     "fp8": 1,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "h100",
    "name": "NVIDIA H100",
    "memoryGB": 80,
    "engine": "vLLM",
    "byFormat": {
     "bf16": 1,
     "fp8": 1,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "h200",
    "name": "NVIDIA H200",
    "memoryGB": 141,
    "engine": "vLLM",
    "byFormat": {
     "bf16": 1,
     "fp8": 1,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "b200",
    "name": "NVIDIA B200 (HGX)",
    "memoryGB": 180,
    "engine": "vLLM",
    "byFormat": {
     "bf16": 1,
     "fp8": 1,
     "q4_k_m": 1
    }
   }
  ]
 },
 "ggufFiles": {
  "repo": "mradermacher/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16-GGUF",
  "gib": {
   "IQ4_XS": 17.66,
   "Q2_K": 17.37,
   "Q3_K_L": 20.14,
   "Q3_K_M": 19.26,
   "Q3_K_S": 17.4,
   "Q4_K_M": 23.68,
   "Q4_K_S": 21.27,
   "Q5_K_M": 25.18,
   "Q5_K_S": 23.11,
   "Q6_K": 32.52,
   "Q8_0": 32.6
  }
 },
 "runsWith": [
  "vLLM",
  "SGLang",
  "llama.cpp (Ollama, LM Studio)",
  "MLX (Mac)"
 ],
 "verified": {
  "gguf": {
   "day": "2026-10-12",
   "repo": "mradermacher/NVIDIA-Nemotron-3.5-Lightning-30B-A3B-BF16-GGUF",
   "kv": "differs",
   "ours": 53248,
   "theirs": 7168
  }
 },
 "updated": "2026-10-12T11:05:05.798Z"
}