NanoJev-Web / evals /runtime-1.1.json
candypunk's picture
Release runtime 1.1 with named workflows and hybrid by default
d786697 verified
Raw History Blame Contribute Delete
4.08 kB
{
"runtimeVersion": "1.1.0",
"weightsUnchanged": true,
"checkpoint": "browser-head-v5",
"hardware": "Apple M3 Max",
"unifiedMemoryGB": 36,
"macOS": "26.5.2",
"node": "24.19.0",
"browser": "agent-browser 0.37.1",
"inference": "local PyTorch MPS FP32",
"runs": [
{
"mode": "model",
"status": "completed",
"elapsedMs": 6997,
"metrics": {
"modelDecisions": 10,
"executedActions": 9,
"deterministicActions": 0,
"blockedRetries": 0,
"browserCalls": 55,
"browserMs": 5114,
"modelMs": 1878,
"inputTokens": 7074,
"outputTokens": 0,
"localApiCostUsd": 0
},
"visibleSuccess": true,
"completedSteps": 6
},
{
"mode": "hybrid",
"status": "completed",
"elapsedMs": 3319,
"metrics": {
"modelDecisions": 5,
"executedActions": 9,
"deterministicActions": 5,
"blockedRetries": 0,
"browserCalls": 45,
"browserMs": 2452,
"modelMs": 864,
"inputTokens": 3175,
"outputTokens": 0,
"localApiCostUsd": 0
},
"visibleSuccess": true,
"completedSteps": 6
},
{
"mode": "hybrid",
"status": "completed",
"elapsedMs": 3358,
"metrics": {
"modelDecisions": 5,
"executedActions": 9,
"deterministicActions": 5,
"blockedRetries": 0,
"browserCalls": 45,
"browserMs": 2488,
"modelMs": 867,
"inputTokens": 3175,
"outputTokens": 0,
"localApiCostUsd": 0
},
"visibleSuccess": true,
"completedSteps": 6
},
{
"mode": "model",
"status": "completed",
"elapsedMs": 4594,
"metrics": {
"modelDecisions": 10,
"executedActions": 9,
"deterministicActions": 0,
"blockedRetries": 0,
"browserCalls": 55,
"browserMs": 2942,
"modelMs": 1650,
"inputTokens": 7074,
"outputTokens": 0,
"localApiCostUsd": 0
},
"visibleSuccess": true,
"completedSteps": 6
},
{
"mode": "model",
"status": "completed",
"elapsedMs": 4640,
"metrics": {
"modelDecisions": 10,
"executedActions": 9,
"deterministicActions": 0,
"blockedRetries": 0,
"browserCalls": 55,
"browserMs": 2970,
"modelMs": 1667,
"inputTokens": 7074,
"outputTokens": 0,
"localApiCostUsd": 0
},
"visibleSuccess": true,
"completedSteps": 6
},
{
"mode": "hybrid",
"status": "completed",
"elapsedMs": 3377,
"metrics": {
"modelDecisions": 5,
"executedActions": 9,
"deterministicActions": 5,
"blockedRetries": 0,
"browserCalls": 45,
"browserMs": 2507,
"modelMs": 867,
"inputTokens": 3175,
"outputTokens": 0,
"localApiCostUsd": 0
},
"visibleSuccess": true,
"completedSteps": 6
}
],
"medianElapsedMs": {
"model": 4640,
"hybrid": 3358
},
"relativeMedianReduction": 0.27629310344827585,
"scope": "Three runs per mode on one synthetic five-field modal workflow; loop timing excludes setup and final evidence. Not a comparison with the original 15-field benchmark or a general reliability claim.",
"checks": {
"attachedSessionPreserved": true,
"compactCliOutput": true,
"independentReceipts": 7,
"exactSavedData": true,
"ambiguousLabelsRejected": true,
"legacyRevealNormalized": true,
"animationObservation": {
"settled": true,
"waitedMs": 267.19999998807907
}
},
"requestContractTests": 6,
"runtimeTests": 10,
"externalModelCalls": 0,
"defaultExecutionMode": "hybrid",
"hybridDefaultValidation": {
"runtimeUnitTestsPassed": 11,
"cliWithoutModeFlagPassed": true,
"assignedInputsExecutedDeterministically": 5,
"exactServerReceipts": 1,
"attachedSessionPreserved": true,
"modelOnlyOverridePassed": true
}
}