{
 "id": "granite-4.1-3b",
 "name": "Granite 4.1 3B",
 "repo": "ibm-granite/granite-4.1-3b",
 "lab": {
  "name": "IBM",
  "country": "United States"
 },
 "released": "2026-04-06",
 "parameters": 3.40283648,
 "activeParameters": null,
 "maxContext": 131072,
 "architecture": "40 layers · 40 full attention (8 KV × 64)",
 "page": "https://studiotvai.com/vram-requirements/granite-4-1-3b",
 "calculator": "https://studiotvai.com/llm-vram-calculator?model=granite-4.1-3b",
 "weightsGiB": {
  "bf16": 6.34,
  "fp8": 3.41,
  "q4_k_m": 1.97
 },
 "kvCacheGiBPerRequest": [
  {
   "context": 8192,
   "bf16": 0.625,
   "fp8": 0.313
  },
  {
   "context": 32768,
   "bf16": 2.5,
   "fp8": 1.25
  },
  {
   "context": 131072,
   "bf16": 10,
   "fp8": 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": 1,
     "fp8": 1,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "rtx-5090",
    "name": "GeForce RTX 5090",
    "memoryGB": 32,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 1,
     "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": "strix-halo",
    "name": "Ryzen AI Max+ 395 128 GB",
    "memoryGB": 96,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 1,
     "fp8": 1,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "dgx-spark",
    "name": "DGX Spark (GB10)",
    "memoryGB": 128,
    "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": "ibm-granite/granite-4.1-3b-GGUF",
  "gib": {
   "Q2_K": 1.28,
   "Q3_K_L": 1.74,
   "Q3_K_M": 1.61,
   "Q3_K_S": 1.46,
   "Q4_0": 1.85,
   "Q4_1": 2.03,
   "Q4_K_M": 1.96,
   "Q4_K_S": 1.86,
   "Q5_0": 2.21,
   "Q5_1": 2.4,
   "Q5_K_M": 2.27,
   "Q5_K_S": 2.21,
   "Q6_K": 2.6,
   "Q8_0": 3.37,
   "BF16": 6.34
  }
 },
 "runsWith": [
  "vLLM",
  "SGLang",
  "llama.cpp (Ollama, LM Studio)",
  "MLX (Mac)"
 ],
 "verified": {
  "gguf": {
   "day": "2026-10-30",
   "repo": "ibm-granite/granite-4.1-3b-GGUF",
   "kv": "match",
   "ours": 81920,
   "theirs": 81920
  }
 },
 "updated": "2026-10-30T00:07:05.915Z"
}