{
 "id": "command-a-plus",
 "name": "Command A+",
 "repo": "CohereLabs/command-a-plus-05-2026-bf16",
 "lab": {
  "name": "Cohere",
  "country": "Canada"
 },
 "released": "2026-05-11",
 "parameters": 218.75,
 "activeParameters": 25,
 "maxContext": 200000,
 "architecture": "MoE 128 experts + shared · 8 global layers (no position embedding) + 24 sliding-window layers (4096 tokens), GQA 8 KV × 128 · vision encoder",
 "page": "https://studiotvai.com/vram-requirements/command-a-plus",
 "calculator": "https://studiotvai.com/llm-vram-calculator?model=command-a-plus",
 "weightsGiB": {
  "bf16": 407.45,
  "fp8": 204.73,
  "q4_k_m": 122.94
 },
 "kvCacheGiBPerRequest": [
  {
   "context": 8192,
   "bf16": 0.625,
   "fp8": 0.313
  },
  {
   "context": 32768,
   "bf16": 1.375,
   "fp8": 0.688
  },
  {
   "context": 131072,
   "bf16": 4.375,
   "fp8": 2.188
  },
  {
   "context": 200000,
   "bf16": 6.479,
   "fp8": 3.239
  }
 ],
 "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": 24,
     "fp8": 16,
     "q4_k_m": 6
    }
   },
   {
    "gpu": "rtx-5090",
    "name": "GeForce RTX 5090",
    "memoryGB": 32,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 16,
     "fp8": 8,
     "q4_k_m": 5
    }
   },
   {
    "gpu": "rtx-pro-6000",
    "name": "RTX PRO 6000 Blackwell",
    "memoryGB": 96,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 5,
     "fp8": 3,
     "q4_k_m": 2
    }
   },
   {
    "gpu": "mac-m5-max-128",
    "name": "Mac M5 Max 128 GB",
    "memoryGB": 96,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 5,
     "fp8": 3,
     "q4_k_m": 2
    }
   },
   {
    "gpu": "strix-halo",
    "name": "Ryzen AI Max+ 395 128 GB",
    "memoryGB": 96,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 5,
     "fp8": 3,
     "q4_k_m": 2
    }
   },
   {
    "gpu": "dgx-spark",
    "name": "DGX Spark (GB10)",
    "memoryGB": 128,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 4,
     "fp8": 2,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "h100",
    "name": "NVIDIA H100",
    "memoryGB": 80,
    "engine": "vLLM",
    "byFormat": {
     "bf16": 7,
     "fp8": 4,
     "q4_k_m": 2
    }
   },
   {
    "gpu": "h200",
    "name": "NVIDIA H200",
    "memoryGB": 141,
    "engine": "vLLM",
    "byFormat": {
     "bf16": 4,
     "fp8": 2,
     "q4_k_m": 2
    }
   },
   {
    "gpu": "b200",
    "name": "NVIDIA B200 (HGX)",
    "memoryGB": 180,
    "engine": "vLLM",
    "byFormat": {
     "bf16": 3,
     "fp8": 2,
     "q4_k_m": 1
    }
   }
  ]
 },
 "ggufFiles": {
  "repo": "DevQuasar/CohereLabs.command-a-plus-05-2026-bf16-GGUF-EXPERIMENTAL",
  "gib": {
   "Q2_K": 74.19,
   "Q3_K_M": 97.36,
   "Q4_K_M": 123.18,
   "Q5_K_M": 144.33,
   "Q6_K": 166.8,
   "Q8_0": 216.03
  }
 },
 "runsWith": [
  "vLLM",
  "SGLang",
  "llama.cpp (Ollama, LM Studio)"
 ],
 "verified": null,
 "updated": "2026-10-30T00:07:05.915Z"
}