{
 "id": "llama-4-maverick",
 "name": "Llama 4 Maverick 17B-128E",
 "repo": "meta-llama/Llama-4-Maverick-17B-128E-Instruct",
 "lab": {
  "name": "Meta",
  "country": "United States"
 },
 "released": "2025-04-01",
 "parameters": 401.58,
 "activeParameters": 17,
 "maxContext": 1048576,
 "architecture": "MoE 128 experts · 36 chunked-attention layers (8192-token chunks, 8 KV × 128) · 12 global layers",
 "page": "https://studiotvai.com/vram-requirements/llama-4-maverick",
 "calculator": "https://studiotvai.com/llm-vram-calculator?model=llama-4-maverick",
 "weightsGiB": {
  "bf16": 748,
  "fp8": 375.93,
  "q4_k_m": 225.52
 },
 "kvCacheGiBPerRequest": [
  {
   "context": 8192,
   "bf16": 1.5,
   "fp8": 0.75
  },
  {
   "context": 32768,
   "bf16": 2.625,
   "fp8": 1.313
  },
  {
   "context": 131072,
   "bf16": 7.125,
   "fp8": 3.563
  },
  {
   "context": 1048576,
   "bf16": 49.125,
   "fp8": 24.563
  }
 ],
 "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": 48,
     "fp8": 24,
     "q4_k_m": 16
    }
   },
   {
    "gpu": "rtx-5090",
    "name": "GeForce RTX 5090",
    "memoryGB": 32,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 32,
     "fp8": 16,
     "q4_k_m": 8
    }
   },
   {
    "gpu": "rtx-pro-6000",
    "name": "RTX PRO 6000 Blackwell",
    "memoryGB": 96,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 16,
     "fp8": 5,
     "q4_k_m": 3
    }
   },
   {
    "gpu": "mac-m5-max-128",
    "name": "Mac M5 Max 128 GB",
    "memoryGB": 96,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 16,
     "fp8": 5,
     "q4_k_m": 3
    }
   },
   {
    "gpu": "h100",
    "name": "NVIDIA H100",
    "memoryGB": 80,
    "engine": "vLLM",
    "byFormat": {
     "bf16": 16,
     "fp8": 6,
     "q4_k_m": 4
    }
   },
   {
    "gpu": "h200",
    "name": "NVIDIA H200",
    "memoryGB": 141,
    "engine": "vLLM",
    "byFormat": {
     "bf16": 7,
     "fp8": 4,
     "q4_k_m": 2
    }
   },
   {
    "gpu": "b200",
    "name": "NVIDIA B200 (HGX)",
    "memoryGB": 180,
    "engine": "vLLM",
    "byFormat": {
     "bf16": 5,
     "fp8": 3,
     "q4_k_m": 2
    }
   }
  ]
 },
 "ggufFiles": {
  "repo": "unsloth/Llama-4-Maverick-17B-128E-Instruct-GGUF",
  "gib": {
   "BF16": 746.43,
   "IQ4_NL": 210.26,
   "IQ4_XS": 199.61,
   "UD-TQ1_0": 98.45,
   "Q2_K": 135.64,
   "Q2_K_L": 135.87,
   "Q3_K_M": 177.95,
   "Q3_K_S": 160.8,
   "Q4_0": 211.19,
   "Q4_1": 233.5,
   "Q4_K_M": 226.09,
   "Q4_K_S": 212.15,
   "Q5_K_M": 264.93,
   "Q5_K_S": 256.76,
   "Q6_K": 306.19,
   "Q8_0": 396.57,
   "UD-IQ1_M": 118.78,
   "UD-IQ1_S": 112.48,
   "UD-IQ2_M": 131.54,
   "UD-IQ2_XXS": 125.91,
   "UD-IQ3_XXS": 157.69,
   "UD-Q2_K_XL": 142.17,
   "UD-Q3_K_XL": 167.22,
   "UD-Q4_K_XL": 216.19,
   "UD-Q5_K_XL": 267.3,
   "UD-Q6_K_XL": 317.65,
   "UD-Q8_K_XL": 428.4
  }
 },
 "runsWith": [
  "vLLM",
  "SGLang",
  "llama.cpp (Ollama, LM Studio)",
  "MLX (Mac)"
 ],
 "verified": null,
 "updated": "2026-10-10T15:05:26.095Z"
}