{
 "id": "step-3.5-flash",
 "name": "Step 3.5 Flash",
 "repo": "stepfun-ai/Step-3.5-Flash",
 "lab": {
  "name": "StepFun",
  "country": "China"
 },
 "released": "2026-02-01",
 "parameters": 199.38,
 "activeParameters": 11,
 "maxContext": 262144,
 "architecture": "MoE 288 experts + 1 shared · 12 full-attention layers + 33 sliding-window layers (512 tokens), GQA 8 KV × 128",
 "page": "https://studiotvai.com/vram-requirements/step-3-5-flash",
 "calculator": "https://studiotvai.com/llm-vram-calculator?model=step-3.5-flash",
 "weightsGiB": {
  "bf16": 366.85,
  "fp8": 184.41,
  "q4_k_m": 110.85
 },
 "kvCacheGiBPerRequest": [
  {
   "context": 8192,
   "bf16": 0.439,
   "fp8": 0.22
  },
  {
   "context": 32768,
   "bf16": 1.564,
   "fp8": 0.782
  },
  {
   "context": 131072,
   "bf16": 6.064,
   "fp8": 3.032
  },
  {
   "context": 262144,
   "bf16": 12.064,
   "fp8": 6.032
  }
 ],
 "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": 7,
     "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": "strix-halo",
    "name": "Ryzen AI Max+ 395 128 GB",
    "memoryGB": 96,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 4,
     "fp8": 2,
     "q4_k_m": 2
    }
   },
   {
    "gpu": "dgx-spark",
    "name": "DGX Spark (GB10)",
    "memoryGB": 128,
    "engine": "llama.cpp",
    "byFormat": {
     "bf16": 3,
     "fp8": 2,
     "q4_k_m": 1
    }
   },
   {
    "gpu": "h100",
    "name": "NVIDIA H100",
    "memoryGB": 80,
    "engine": "vLLM",
    "byFormat": {
     "bf16": 6,
     "fp8": 3,
     "q4_k_m": 2
    }
   },
   {
    "gpu": "h200",
    "name": "NVIDIA H200",
    "memoryGB": 141,
    "engine": "vLLM",
    "byFormat": {
     "bf16": 4,
     "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": [],
 "verified": null,
 "updated": "2026-10-16T07:08:10.630Z"
}