Google
Unsloth

Gemma 4 E2B IT benchmark on an Intel's logo.Arc Graphics

<- Runs

Prompt tokens

40,960

Generation tokens

10,240

Trials passed

10/10

Verified

8.7 tok/s

116.9 tok/s

Peak memory

4.97/31 GB

Runs poorly

Trials

Decode / Prefill Speeds

Metadata

metadata.json
{
"runId": "run_c6905051-2253-4934-acbc-cab81e6a44e1",
"bundleId": "llamacpp-gemma-4-e4b-it-ud-q4_k_xl.gguf-108968",
"status": "verified",
"promptTokens": 40960,
"completionTokens": 10240,
"contextLength": 5120,
"harness": {
"version": "0.1.21",
"gitSha": "05c5778"
},
"runtime": {
"name": "llama.cpp",
"version": "b9010",
"buildFlags": "metal"
},
"model": {
"displayName": "Gemma 4 E2B IT",
"format": "gguf",
"quant": "ud_q4_k_xl",
"architecture": "gemma4",
"source": "unsloth/gemma-4-E4B-it-GGUF:gemma-4-E4B-it-UD-Q4_K_XL.gguf",
"fileSizeBytes": 5126304928,
"lab": {
"name": "Google",
"slug": "google"
},
"quantizedBy": {
"name": "Unsloth",
"slug": "unsloth"
}
},
"device": {
"cpu": "Intel(R) Core(TM) Ultra 7 155H",
"cpuCores": 22,
"gpu": "Intel Arc Graphics",
"gpuCores": 0,
"gpuCount": 1,
"ramGb": 31,
"osName": "HamoniKR 8.0",
"osVersion": "8.0"
},
"decodeTpsMean": 8.7,
"prefillTpsMean": 116.9,
"ttftP50Ms": 34107.18,
"idleTpsMean": 2501.3,
"peakRssMb": 5086.2,
"trialsPassed": 10,
"trialsTotal": 10,
"runnabilityScore": 0.4171569052419355,
"bundleSha256": "499196a41c8a8d62f3bb6e173136f279d143f4dc184c55a5209fd5d1aa40d886",
"createdAt": "2026-05-05T01:35:05.975Z"
}