Google
LM Studio

Gemma 4 26B A4B IT benchmark on a NVIDIA's logo.GeForce RTX 3060

<- Runs

Prompt tokens

40,960

Generation tokens

10,240

Trials passed

10/10

Verified

7.3 tok/s

125.3 tok/s

Peak memory

14.18/16 GB

Barely runs

Trials

Decode / Prefill Speeds

Metadata

metadata.json
{
"runId": "run_6b022f7d-0594-4e5d-8ce9-ad4a190a4406",
"bundleId": "llamacpp-gemma-4-26b-a4b-it-q4_k_m.gguf-f3978c",
"status": "verified",
"promptTokens": 40960,
"completionTokens": 10240,
"contextLength": 5120,
"harness": {
"version": "0.1.21",
"gitSha": "unknown"
},
"runtime": {
"name": "llama.cpp",
"version": "b8796",
"buildFlags": "metal"
},
"model": {
"displayName": "Gemma 4 26B A4B IT",
"format": "gguf",
"quant": "q4_k_m",
"architecture": "gemma4",
"source": "lmstudio-community/gemma-4-26B-A4B-it-GGUF:gemma-4-26B-A4B-it-Q4_K_M.gguf",
"fileSizeBytes": 16796015040,
"lab": {
"name": "Google",
"slug": "google"
},
"quantizedBy": {
"name": "LM Studio",
"slug": "lmstudio"
}
},
"device": {
"cpu": "12th Gen Intel(R) Core(TM) i7-12700F",
"cpuCores": 20,
"gpu": "NVIDIA GeForce RTX 3060",
"gpuCores": 0,
"gpuCount": 1,
"ramGb": 16,
"osName": "Ubuntu 22.04.2 LTS",
"osVersion": "22.04"
},
"decodeTpsMean": 7.3,
"prefillTpsMean": 125.3,
"ttftP50Ms": 32676.25,
"idleTpsMean": 14029.2,
"peakRssMb": 14523.4,
"trialsPassed": 10,
"trialsTotal": 10,
"runnabilityScore": 0.162725,
"bundleSha256": "175b4799fad5802200117c2afd852dbcbeedacb5a44f58fe06708d68638edb14",
"createdAt": "2026-04-18T06:36:14.604Z"
}