{
 "id": "agnes-3.0-qwen",
 "name": "Agnes-3.0-Qwen",
 "repo": "Agnes-AI/Agnes-3.0-Qwen",
 "lab": {
  "name": "Agnes AI",
  "country": "Singapore"
 },
 "released": "2026-10-10",
 "parameters": 179.999981459,
 "activeParameters": 7.8,
 "maxContext": 1048576,
 "architecture": "48 layers · 12 full attention (2 KV × 256) · indexer keys per 4-token block · 36 linear attention",
 "page": "https://studiotvai.com/vram-requirements/agnes-3-0-qwen",
 "calculator": "https://studiotvai.com/llm-vram-calculator?model=agnes-3.0-qwen",
 "weightsGiB": {
  "bf16": 330.42,
  "fp8": 166.39,
  "q4_k_m": 105.07
 },
 "kvCacheGiBPerRequest": [
  {
   "context": 8192,
   "bf16": 0.193,
   "fp8": 0.1
  },
  {
   "context": 32768,
   "bf16": 0.773,
   "fp8": 0.398
  },
  {
   "context": 131072,
   "bf16": 3.094,
   "fp8": 1.594
  },
  {
   "context": 1048576,
   "bf16": 24.75,
   "fp8": 12.75
  }
 ],
 "recurrentStateGiBPerRequest": 0.1075,
 "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": 16,
     "fp8": 8,
     "q4_k_m": 5
    }
   },
   {
    "gpu": "rtx-5090",
    "name": "GeForce RTX 5090",
    "memoryGB": 32,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 16,
     "fp8": 6,
     "q4_k_m": 4
    }
   },
   {
    "gpu": "rtx-pro-6000",
    "name": "RTX PRO 6000 Blackwell",
    "memoryGB": 96,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 4,
     "fp8": 2,
     "q4_k_m": 2
    }
   },
   {
    "gpu": "mac-m5-max-128",
    "name": "Mac M5 Max 128 GB",
    "memoryGB": 96,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 4,
     "fp8": 2,
     "q4_k_m": 2
    }
   },
   {
    "gpu": "h100",
    "name": "NVIDIA H100",
    "memoryGB": 80,
    "engine": "vLLM",
    "byFormat": {
     "bf16": 5,
     "fp8": 3,
     "q4_k_m": 2
    }
   },
   {
    "gpu": "h200",
    "name": "NVIDIA H200",
    "memoryGB": 141,
    "engine": "vLLM",
    "byFormat": {
     "bf16": 3,
     "fp8": 2,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "b200",
    "name": "NVIDIA B200 (HGX)",
    "memoryGB": 180,
    "engine": "vLLM",
    "byFormat": {
     "bf16": 3,
     "fp8": 2,
     "q4_k_m": 1
    }
   }
  ]
 },
 "ggufFiles": null,
 "runsWith": [
  "vLLM",
  "SGLang",
  "llama.cpp (Ollama, LM Studio)"
 ],
 "verified": null,
 "updated": "2026-10-10T15:05:26.095Z"
}