{ "protocolVersion": "osai-bench/1.3", "clientVersion": "0.10.0", "scoringVersion": "osai-bench-derive/1.3", "createdAt": "2026-08-03T01:32:14.094Z", "qualityOverride": false, "cohortEligible": true, "qualityConditions": [], "runtime": { "name": "ollama", "version": "0.32.5", "endpoint": "loopback", "layerAssignment": null, "offloadPlacement": { "source": "ollama.ps.size_vram", "granularity": "bytes", "residentBytes": 21519229313, "vramResidentBytes": 7808714014, "hostResidentBytes": 13710515299, "vramResidentFraction": 0.3628714532672725 }, "kvCacheMetadata": { "source": "ollama.show.model_info", "architecture": "gemma4", "blockCount": 60, "kvHeadCount": null, "attentionHeadCount": 32, "embeddingLength": 5376, "headDimension": 512, "resolvedElementType": null, "projectedBytes": null, "calculationAvailable": false, "missingInputs": [ "kvHeadCount", "resolvedKvCacheElementType" ] }, "environment": { "source": "client-process-env", "authoritative": false, "note": "Read from the benchmark client's own process environment. The Ollama server may have been started with a different environment; Ollama exposes no endpoint reporting its resolved configuration.", "declared": { "OLLAMA_KV_CACHE_TYPE": null, "OLLAMA_FLASH_ATTENTION": null, "OLLAMA_CONTEXT_LENGTH": null, "OLLAMA_NUM_PARALLEL": null, "OLLAMA_GPU_OVERHEAD": null, "OLLAMA_SCHED_SPREAD": null, "OLLAMA_LLM_LIBRARY": null, "LLAMA_ARG_FIT": null, "LLAMA_ARG_FIT_TARGET": null, "OLLAMA_MAX_LOADED_MODELS": null, "OLLAMA_KEEP_ALIVE": null, "OLLAMA_DEBUG": null, "OLLAMA_MODELS": false, "OLLAMA_HOST": false }, "declaredNonDefault": [] } }, "model": { "identifier": "gemma4:31b", "digest": "6316f0629137b426c9d9b853ffc4c8209589f30ee39aebede6285096c0ff47e7", "family": "gemma4", "parameterSize": "31.3B", "quantization": "Q4_K_M", "weightsBytes": 19868981791, "weightsSource": "ollama.tags.size" }, "system": { "cpu": { "model": "12th Gen Intel(R) Core(TM) i9-12900K" }, "gpu": { "present": true, "model": "NVIDIA GeForce RTX 3080", "totalVramBytes": 10737418240, "freeVramBytesAtCheck": 10326376448, "freeVramBytesAtLoad": 963641344, "utilizationPercentAtCheck": 0, "driverVersion": "595.71.05", "provider": "nvidia-smi" }, "memory": { "totalBytes": 32817025024 }, "os": { "platform": "linux", "version": "Ubuntu 26.04 LTS", "architecture": "x64" }, "power": { "present": false, "onBattery": false } }, "configuration": { "memoryBandwidthGBps": 760, "memoryBandwidthSource": "manufacturer-table", "memoryBandwidthTableVersion": "osai-gpu-memory-bandwidth/2", "memoryBandwidthEntryId": "nvidia-geforce-rtx-3080-10gb", "fixedOptions": { "temperature": 0, "seed": 42, "stream": true }, "workloads": { "w1": { "numPredict": 1, "numCtx": 512, "keepAlive": "5m", "promptTokenRange": null, "warmups": 0, "repetitions": 1, "varyPromptPerCall": false }, "w2": { "numPredict": 128, "numCtx": 4096, "keepAlive": "5m", "promptTokenRange": { "min": 20, "max": 64 }, "warmups": 1, "repetitions": 5, "varyPromptPerCall": false }, "w3": { "numPredict": 1, "numCtx": 4096, "keepAlive": "5m", "promptTokenRange": { "min": 2000, "max": 4095 }, "warmups": 1, "repetitions": 5, "varyPromptPerCall": true }, "w4": { "numPredict": 512, "numCtx": 4096, "keepAlive": "5m", "promptTokenRange": { "min": 20, "max": 64 }, "warmups": 1, "repetitions": 5, "varyPromptPerCall": false } }, "resolved": { "fixedOptions": { "temperature": 0, "seed": 42, "stream": true }, "workloads": { "w1": { "numPredict": 1, "numCtx": 512, "keepAlive": "5m" }, "w2": { "numPredict": 128, "numCtx": 4096, "keepAlive": "5m" }, "w3": { "numPredict": 1, "numCtx": 4096, "keepAlive": "5m" }, "w4": { "numPredict": 512, "numCtx": 4096, "keepAlive": "5m" } }, "runtimeModelParameters": "temperature 1\ntop_k 64\ntop_p 0.95", "unresolvedRuntimeDefaults": true } }, "rawMeasurements": { "workloads": { "w1": { "warmup": null, "measuredPasses": [ { "index": 1, "valid": true, "attempts": [ { "measurement": { "total_duration": 8902926677, "load_duration": 7042170816, "prompt_eval_count": 21, "prompt_eval_duration": 1858861000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 8902926677, "load_duration": 7042170816, "prompt_eval_count": 21, "prompt_eval_duration": 1858861000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "failed": false }, "w2": { "warmup": { "total_duration": 62778728376, "load_duration": 6976081284, "prompt_eval_count": 55, "prompt_eval_duration": 2370954000, "eval_count": 128, "eval_duration": 53391961000, "timeToFirstTokenMs": null }, "measuredPasses": [ { "index": 1, "valid": true, "attempts": [ { "measurement": { "total_duration": 54763195197, "load_duration": 316808532, "prompt_eval_count": 55, "prompt_eval_duration": 936831000, "eval_count": 128, "eval_duration": 53429126000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 54763195197, "load_duration": 316808532, "prompt_eval_count": 55, "prompt_eval_duration": 936831000, "eval_count": 128, "eval_duration": 53429126000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 2, "valid": true, "attempts": [ { "measurement": { "total_duration": 54822263163, "load_duration": 384815502, "prompt_eval_count": 55, "prompt_eval_duration": 935330000, "eval_count": 128, "eval_duration": 53398967000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 54822263163, "load_duration": 384815502, "prompt_eval_count": 55, "prompt_eval_duration": 935330000, "eval_count": 128, "eval_duration": 53398967000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 3, "valid": true, "attempts": [ { "measurement": { "total_duration": 54679078435, "load_duration": 304395811, "prompt_eval_count": 55, "prompt_eval_duration": 931114000, "eval_count": 128, "eval_duration": 53441771000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 54679078435, "load_duration": 304395811, "prompt_eval_count": 55, "prompt_eval_duration": 931114000, "eval_count": 128, "eval_duration": 53441771000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 4, "valid": true, "attempts": [ { "measurement": { "total_duration": 54674135144, "load_duration": 325164328, "prompt_eval_count": 55, "prompt_eval_duration": 939255000, "eval_count": 128, "eval_duration": 53407087000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 54674135144, "load_duration": 325164328, "prompt_eval_count": 55, "prompt_eval_duration": 939255000, "eval_count": 128, "eval_duration": 53407087000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 5, "valid": true, "attempts": [ { "measurement": { "total_duration": 54651924922, "load_duration": 318471096, "prompt_eval_count": 55, "prompt_eval_duration": 935931000, "eval_count": 128, "eval_duration": 53395701000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 54651924922, "load_duration": 318471096, "prompt_eval_count": 55, "prompt_eval_duration": 935931000, "eval_count": 128, "eval_duration": 53395701000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "failed": false }, "w3": { "warmup": { "total_duration": 15392768062, "load_duration": 318285433, "prompt_eval_count": 2681, "prompt_eval_duration": 15055112000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "measuredPasses": [ { "index": 1, "valid": true, "attempts": [ { "measurement": { "total_duration": 16291447736, "load_duration": 312854397, "prompt_eval_count": 2681, "prompt_eval_duration": 15079019000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 16291447736, "load_duration": 312854397, "prompt_eval_count": 2681, "prompt_eval_duration": 15079019000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 2, "valid": true, "attempts": [ { "measurement": { "total_duration": 16793255571, "load_duration": 314645430, "prompt_eval_count": 2681, "prompt_eval_duration": 15126131000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 16793255571, "load_duration": 314645430, "prompt_eval_count": 2681, "prompt_eval_duration": 15126131000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 3, "valid": true, "attempts": [ { "measurement": { "total_duration": 17317232944, "load_duration": 351872314, "prompt_eval_count": 2681, "prompt_eval_duration": 15093072000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 17317232944, "load_duration": 351872314, "prompt_eval_count": 2681, "prompt_eval_duration": 15093072000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 4, "valid": true, "attempts": [ { "measurement": { "total_duration": 17415805048, "load_duration": 349982460, "prompt_eval_count": 2681, "prompt_eval_duration": 15184230000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 17415805048, "load_duration": 349982460, "prompt_eval_count": 2681, "prompt_eval_duration": 15184230000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 5, "valid": true, "attempts": [ { "measurement": { "total_duration": 16376339051, "load_duration": 366432087, "prompt_eval_count": 2681, "prompt_eval_duration": 15048499000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 16376339051, "load_duration": 366432087, "prompt_eval_count": 2681, "prompt_eval_duration": 15048499000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "failed": false }, "w4": { "warmup": { "total_duration": 222893077529, "load_duration": 344882121, "prompt_eval_count": 55, "prompt_eval_duration": 2357800000, "eval_count": 512, "eval_duration": 219211092000, "timeToFirstTokenMs": 195708.65972000005 }, "measuredPasses": [ { "index": 1, "valid": true, "attempts": [ { "measurement": { "total_duration": 220705127040, "load_duration": 354893512, "prompt_eval_count": 55, "prompt_eval_duration": 936563000, "eval_count": 512, "eval_duration": 219120407000, "timeToFirstTokenMs": 142468.910761 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 220705127040, "load_duration": 354893512, "prompt_eval_count": 55, "prompt_eval_duration": 936563000, "eval_count": 512, "eval_duration": 219120407000, "timeToFirstTokenMs": 142468.910761 }, "validity": { "valid": true, "reasons": [] } }, { "index": 2, "valid": true, "attempts": [ { "measurement": { "total_duration": 220651176985, "load_duration": 329228476, "prompt_eval_count": 55, "prompt_eval_duration": 940923000, "eval_count": 512, "eval_duration": 219141628000, "timeToFirstTokenMs": 142390.41373399994 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 220651176985, "load_duration": 329228476, "prompt_eval_count": 55, "prompt_eval_duration": 940923000, "eval_count": 512, "eval_duration": 219141628000, "timeToFirstTokenMs": 142390.41373399994 }, "validity": { "valid": true, "reasons": [] } }, { "index": 3, "valid": true, "attempts": [ { "measurement": { "total_duration": 220438790246, "load_duration": 327966165, "prompt_eval_count": 55, "prompt_eval_duration": 938578000, "eval_count": 512, "eval_duration": 219170113000, "timeToFirstTokenMs": 142173.228136 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 220438790246, "load_duration": 327966165, "prompt_eval_count": 55, "prompt_eval_duration": 938578000, "eval_count": 512, "eval_duration": 219170113000, "timeToFirstTokenMs": 142173.228136 }, "validity": { "valid": true, "reasons": [] } }, { "index": 4, "valid": true, "attempts": [ { "measurement": { "total_duration": 220378663260, "load_duration": 312784583, "prompt_eval_count": 55, "prompt_eval_duration": 943889000, "eval_count": 512, "eval_duration": 219118756000, "timeToFirstTokenMs": 142131.80454499996 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 220378663260, "load_duration": 312784583, "prompt_eval_count": 55, "prompt_eval_duration": 943889000, "eval_count": 512, "eval_duration": 219118756000, "timeToFirstTokenMs": 142131.80454499996 }, "validity": { "valid": true, "reasons": [] } }, { "index": 5, "valid": true, "attempts": [ { "measurement": { "total_duration": 220415218793, "load_duration": 337802872, "prompt_eval_count": 55, "prompt_eval_duration": 942310000, "eval_count": 512, "eval_duration": 219132938000, "timeToFirstTokenMs": 142162.638884 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 220415218793, "load_duration": 337802872, "prompt_eval_count": 55, "prompt_eval_duration": 942310000, "eval_count": 512, "eval_duration": 219132938000, "timeToFirstTokenMs": 142162.638884 }, "validity": { "valid": true, "reasons": [] } } ], "failed": false } } }, "derived": { "generationTokensPerSecond": { "median": 2.336481245918402, "coefficientOfVariation": 0.00009523237525563538, "samples": 5 }, "prefillTokensPerSecond": { "median": 177.63116746544375, "coefficientOfVariation": 0.003422384001599265, "samples": 5 }, "timeToFirstTokenMs": { "median": null, "coefficientOfVariation": null, "samples": 5 }, "coldLoad": { "seconds": 7.042170816, "samples": 1 }, "passFailureRate": { "failedMeasuredPasses": 0, "totalMeasuredPasses": 16, "percent": 0 }, "attemptFailureRate": { "failedAttempts": 0, "totalAttempts": 16, "percent": 0 }, "roofline": { "modelWeightsGB": 19.868981791, "memoryBandwidthGBps": 760, "theoreticalMaxTokensPerSecond": 38.25057609868339, "utilization": 0.06108355701337595 }, "diagnostics": [ { "id": "partial-cpu-offload", "status": "detected", "message": "64% of the model's resident bytes are on the host, not in VRAM", "evidence": { "source": "ollama.ps.size_vram", "granularity": "bytes", "residentBytes": 21519229313, "vramResidentBytes": 7808714014, "hostResidentBytes": 13710515299, "vramResidentFraction": 0.3628714532672725 } }, { "id": "context-vram-headroom", "status": "unavailable", "message": "Ollama /api/show does not expose the resolved KV-cache element type actually in use, so v1.1 does not calculate or threshold projected KV-cache VRAM", "evidence": { "configuredContext": 4096, "availableArchitectureMetadata": { "source": "ollama.show.model_info", "architecture": "gemma4", "blockCount": 60, "kvHeadCount": null, "attentionHeadCount": 32, "embeddingLength": 5376, "headDimension": 512, "resolvedElementType": null, "projectedBytes": null, "calculationAvailable": false, "missingInputs": [ "kvHeadCount", "resolvedKvCacheElementType" ] }, "missingInput": "resolvedKvCacheElementType" } }, { "id": "cpu-only-with-gpu", "status": "not-detected", "message": "36% of the model's resident bytes are in VRAM", "evidence": { "source": "ollama.ps.size_vram", "granularity": "bytes", "residentBytes": 21519229313, "vramResidentBytes": 7808714014, "hostResidentBytes": 13710515299, "vramResidentFraction": 0.3628714532672725, "gpuModel": "NVIDIA GeForce RTX 3080" } }, { "id": "weights-exceed-vram", "status": "detected", "message": "Quantized on-disk model weight size is larger than total VRAM", "evidence": { "weightsBytes": 19868981791, "totalVramBytes": 10737418240 } } ] } }