Workload cost report

Hot Aisle

Qwen/Qwen3-Coder-30B-A3B-Instruct

$0.2419

per 1,000 completed requests

Result rate: 3.433 / second
Completed / attempted: 95 / 96
Trials / total measured seconds: 2 / 27.674
Allocation: 1 GPUs × $2.99/hour + $0.0000/hour extras
Cost basis: modeled charge for summed benchmark windows; $0.0230
Price source / period: local demo (fake API on-demand price) / synthetic
Quote / cost note: None

Successful-request latency (ms)

	p50	p95	p99
TTFT	252.6	274.84	278.342
pooled successful-request samples
E2E	8,476.9	9,951.01	10,162.832
pooled successful-request samples

Conditions and input commitments

{
  "identity": {
    "model": "Qwen/Qwen3-Coder-30B-A3B-Instruct",
    "revision": "demo-rev",
    "precision": "FP8",
    "tokenizer": "demo-rev",
    "workload": "sha256:b0ddd66d9e8eb6f903fc0a37240e7da8",
    "cache": "warm",
    "load": "dataset=random;in=2048;out=256;prompts=48;concurrency=32;rate=inf;seed=7",
    "runtime": "rocm/vllm@sha256:demo"
  },
  "gates": {
    "ttft_ms": null,
    "e2e_ms": null,
    "include_client_queue": false,
    "quality": false
  },
  "criterion": "",
  "holds": [],
  "warnings": []
}

Input commitments

SHA-256 724ff34cd45dbd4e55674fcc24ecd87b48be7c7bf019305bd02c50210adfb5d9
Record 0 · 3342 bytes

SHA-256 3f233d4369eb65cea4f48a9cd2877162fe74f8de1e78af9c5a03c7791ec78f93
Record 0 · 3335 bytes