{
 "id": "mistral-small-4",
 "name": "Mistral Small 4 119B",
 "repo": "mistralai/Mistral-Small-4-119B-2603",
 "lab": {
  "name": "Mistral AI",
  "country": "France"
 },
 "released": "2026-03-16",
 "parameters": 119.4,
 "activeParameters": 6.5,
 "maxContext": 262144,
 "architecture": "MoE 128 experts + 1 shared · MLA (256 + 64 latent) on 36 layers · vision encoder",
 "page": "https://studiotvai.com/vram-requirements/mistral-small-4",
 "calculator": "https://studiotvai.com/llm-vram-calculator?model=mistral-small-4",
 "weightsGiB": {
  "native": 112.62,
  "bf16": 222.4,
  "fp8": 112.2,
  "q4_k_m": 67
 },
 "kvCacheGiBPerRequest": [
  {
   "context": 8192,
   "bf16": 0.176,
   "fp8": 0.088
  },
  {
   "context": 32768,
   "bf16": 0.703,
   "fp8": 0.352
  },
  {
   "context": 131072,
   "bf16": 2.813,
   "fp8": 1.406
  },
  {
   "context": 262144,
   "bf16": 5.625,
   "fp8": 2.813
  }
 ],
 "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": {
     "native": 6,
     "bf16": 16,
     "fp8": 6,
     "q4_k_m": 4
    }
   },
   {
    "gpu": "rtx-5090",
    "name": "GeForce RTX 5090",
    "memoryGB": 32,
    "engine": "llama.cpp",
    "byFormat": {
     "native": 4,
     "bf16": 8,
     "fp8": 4,
     "q4_k_m": 3
    }
   },
   {
    "gpu": "rtx-pro-6000",
    "name": "RTX PRO 6000 Blackwell",
    "memoryGB": 96,
    "engine": "llama.cpp",
    "byFormat": {
     "native": 2,
     "bf16": 3,
     "fp8": 2,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "mac-m5-max-128",
    "name": "Mac M5 Max 128 GB",
    "memoryGB": 96,
    "engine": "llama.cpp",
    "byFormat": {
     "native": 2,
     "bf16": 3,
     "fp8": 2,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "strix-halo",
    "name": "Ryzen AI Max+ 395 128 GB",
    "memoryGB": 96,
    "engine": "llama.cpp",
    "byFormat": {
     "native": 2,
     "bf16": 3,
     "fp8": 2,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "dgx-spark",
    "name": "DGX Spark (GB10)",
    "memoryGB": 128,
    "engine": "llama.cpp",
    "byFormat": {
     "native": 1,
     "bf16": 2,
     "fp8": 1,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "h100",
    "name": "NVIDIA H100",
    "memoryGB": 80,
    "engine": "vLLM",
    "byFormat": {
     "native": 2,
     "bf16": 4,
     "fp8": 2,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "h200",
    "name": "NVIDIA H200",
    "memoryGB": 141,
    "engine": "vLLM",
    "byFormat": {
     "native": 1,
     "bf16": 2,
     "fp8": 1,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "b200",
    "name": "NVIDIA B200 (HGX)",
    "memoryGB": 180,
    "engine": "vLLM",
    "byFormat": {
     "native": 1,
     "bf16": 2,
     "fp8": 1,
     "q4_k_m": 1
    }
   }
  ]
 },
 "ggufFiles": {
  "repo": "bartowski/mistralai_Mistral-Small-4-119B-2603-GGUF",
  "gib": {
   "IQ1_M": 27.51,
   "IQ1_S": 25.04,
   "IQ2_M": 38.77,
   "IQ2_S": 35.52,
   "IQ2_XS": 34.48,
   "IQ2_XXS": 31.35,
   "IQ3_M": 53.05,
   "IQ3_XS": 48.2,
   "IQ3_XXS": 46.44,
   "IQ4_NL": 62.83,
   "IQ4_XS": 59.41,
   "Q2_K": 39.18,
   "Q2_K_L": 39.67,
   "Q3_K_L": 52.99,
   "Q3_K_M": 50.82,
   "Q3_K_S": 48.42,
   "Q3_K_XL": 53.43,
   "Q4_0": 63.06,
   "Q4_K_L": 68.02,
   "Q4_K_M": 67.65,
   "Q4_K_S": 65.08,
   "Q5_K_L": 79.51,
   "Q5_K_M": 79.2,
   "Q5_K_S": 76.57,
   "Q6_K": 95.75,
   "Q6_K_L": 95.99,
   "Q8_0": 117.79
  }
 },
 "runsWith": [
  "vLLM",
  "SGLang",
  "llama.cpp (Ollama, LM Studio)",
  "MLX (Mac)"
 ],
 "verified": null,
 "updated": "2026-10-30T00:07:05.915Z"
}