{ "protocolVersion": "osai-bench/1.3", "clientVersion": "0.10.0", "scoringVersion": "osai-bench-derive/1.3", "createdAt": "2026-08-03T06:30:51.837Z", "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": 18814077917, "vramResidentBytes": 7806124031, "hostResidentBytes": 11007953886, "vramResidentFraction": 0.41490866921235364 }, "kvCacheMetadata": { "source": "ollama.show.model_info", "architecture": "gemma4", "blockCount": 30, "kvHeadCount": null, "attentionHeadCount": 16, "embeddingLength": 2816, "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": "gemma4:26b", "digest": "5571076f3d70050487b26b341705799e0ab29b808164f90d20d4cf84f699d251", "family": "gemma4", "parameterSize": "25.8B", "quantization": "Q4_K_M", "weightsBytes": 17987581215, "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": 993001472, "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": 12340869942, "load_duration": 11965959372, "prompt_eval_count": 21, "prompt_eval_duration": 372586000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 12340869942, "load_duration": 11965959372, "prompt_eval_count": 21, "prompt_eval_duration": 372586000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "failed": false }, "w2": { "warmup": { "total_duration": 18335995991, "load_duration": 11959482624, "prompt_eval_count": 55, "prompt_eval_duration": 1979079000, "eval_count": 128, "eval_duration": 4394348000, "timeToFirstTokenMs": null }, "measuredPasses": [ { "index": 1, "valid": true, "attempts": [ { "measurement": { "total_duration": 4931167712, "load_duration": 326144071, "prompt_eval_count": 55, "prompt_eval_duration": 129155000, "eval_count": 128, "eval_duration": 4432350000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 4931167712, "load_duration": 326144071, "prompt_eval_count": 55, "prompt_eval_duration": 129155000, "eval_count": 128, "eval_duration": 4432350000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 2, "valid": true, "attempts": [ { "measurement": { "total_duration": 4924385572, "load_duration": 317817059, "prompt_eval_count": 55, "prompt_eval_duration": 124978000, "eval_count": 128, "eval_duration": 4427411000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 4924385572, "load_duration": 317817059, "prompt_eval_count": 55, "prompt_eval_duration": 124978000, "eval_count": 128, "eval_duration": 4427411000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 3, "valid": true, "attempts": [ { "measurement": { "total_duration": 4869772168, "load_duration": 316653974, "prompt_eval_count": 55, "prompt_eval_duration": 122526000, "eval_count": 128, "eval_duration": 4428693000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 4869772168, "load_duration": 316653974, "prompt_eval_count": 55, "prompt_eval_duration": 122526000, "eval_count": 128, "eval_duration": 4428693000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 4, "valid": true, "attempts": [ { "measurement": { "total_duration": 4904239556, "load_duration": 327554017, "prompt_eval_count": 55, "prompt_eval_duration": 129399000, "eval_count": 128, "eval_duration": 4445019000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 4904239556, "load_duration": 327554017, "prompt_eval_count": 55, "prompt_eval_duration": 129399000, "eval_count": 128, "eval_duration": 4445019000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 5, "valid": true, "attempts": [ { "measurement": { "total_duration": 4896655955, "load_duration": 318383117, "prompt_eval_count": 55, "prompt_eval_duration": 119837000, "eval_count": 128, "eval_duration": 4456407000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 4896655955, "load_duration": 318383117, "prompt_eval_count": 55, "prompt_eval_duration": 119837000, "eval_count": 128, "eval_duration": 4456407000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "failed": false }, "w3": { "warmup": { "total_duration": 14588764112, "load_duration": 311038371, "prompt_eval_count": 2681, "prompt_eval_duration": 14250501000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "measuredPasses": [ { "index": 1, "valid": true, "attempts": [ { "measurement": { "total_duration": 14877527782, "load_duration": 342109488, "prompt_eval_count": 2681, "prompt_eval_duration": 14148494000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 14877527782, "load_duration": 342109488, "prompt_eval_count": 2681, "prompt_eval_duration": 14148494000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 2, "valid": true, "attempts": [ { "measurement": { "total_duration": 14826327472, "load_duration": 309598724, "prompt_eval_count": 2681, "prompt_eval_duration": 14134849000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 14826327472, "load_duration": 309598724, "prompt_eval_count": 2681, "prompt_eval_duration": 14134849000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 3, "valid": true, "attempts": [ { "measurement": { "total_duration": 14870034844, "load_duration": 301857306, "prompt_eval_count": 2681, "prompt_eval_duration": 14193444000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 14870034844, "load_duration": 301857306, "prompt_eval_count": 2681, "prompt_eval_duration": 14193444000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 4, "valid": true, "attempts": [ { "measurement": { "total_duration": 14877743499, "load_duration": 346151077, "prompt_eval_count": 2681, "prompt_eval_duration": 14147656000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 14877743499, "load_duration": 346151077, "prompt_eval_count": 2681, "prompt_eval_duration": 14147656000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 5, "valid": true, "attempts": [ { "measurement": { "total_duration": 14828932120, "load_duration": 313554713, "prompt_eval_count": 2681, "prompt_eval_duration": 14139877000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 14828932120, "load_duration": 313554713, "prompt_eval_count": 2681, "prompt_eval_duration": 14139877000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "failed": false }, "w4": { "warmup": { "total_duration": 17986881026, "load_duration": 310445328, "prompt_eval_count": 55, "prompt_eval_duration": 131651999, "eval_count": 512, "eval_duration": 17147367999, "timeToFirstTokenMs": null }, "measuredPasses": [ { "index": 1, "valid": true, "attempts": [ { "measurement": { "total_duration": 17735218504, "load_duration": 320416869, "prompt_eval_count": 55, "prompt_eval_duration": 131384999, "eval_count": 512, "eval_duration": 17169114000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 17735218504, "load_duration": 320416869, "prompt_eval_count": 55, "prompt_eval_duration": 131384999, "eval_count": 512, "eval_duration": 17169114000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 2, "valid": true, "attempts": [ { "measurement": { "total_duration": 17666542570, "load_duration": 342921420, "prompt_eval_count": 55, "prompt_eval_duration": 118800000, "eval_count": 512, "eval_duration": 17203026000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 17666542570, "load_duration": 342921420, "prompt_eval_count": 55, "prompt_eval_duration": 118800000, "eval_count": 512, "eval_duration": 17203026000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 3, "valid": true, "attempts": [ { "measurement": { "total_duration": 17626882875, "load_duration": 324683129, "prompt_eval_count": 55, "prompt_eval_duration": 117988000, "eval_count": 512, "eval_duration": 17182274000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 17626882875, "load_duration": 324683129, "prompt_eval_count": 55, "prompt_eval_duration": 117988000, "eval_count": 512, "eval_duration": 17182274000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 4, "valid": true, "attempts": [ { "measurement": { "total_duration": 17582769290, "load_duration": 303594214, "prompt_eval_count": 55, "prompt_eval_duration": 118006000, "eval_count": 512, "eval_duration": 17159283000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 17582769290, "load_duration": 303594214, "prompt_eval_count": 55, "prompt_eval_duration": 118006000, "eval_count": 512, "eval_duration": 17159283000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } }, { "index": 5, "valid": true, "attempts": [ { "measurement": { "total_duration": 17604449495, "load_duration": 331269652, "prompt_eval_count": 55, "prompt_eval_duration": 118105000, "eval_count": 512, "eval_duration": 17153032000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 17604449495, "load_duration": 331269652, "prompt_eval_count": 55, "prompt_eval_duration": 118105000, "eval_count": 512, "eval_duration": 17153032000, "timeToFirstTokenMs": null }, "validity": { "valid": true, "reasons": [] } } ], "failed": false } } }, "derived": { "generationTokensPerSecond": { "median": 29.820991345272677, "coefficientOfVariation": 0.0011599035943480904, "samples": 5 }, "prefillTokensPerSecond": { "median": 189.5013562670735, "coefficientOfVariation": 0.0016487157199733683, "samples": 5 }, "timeToFirstTokenMs": { "median": null, "coefficientOfVariation": null, "samples": 5 }, "coldLoad": { "seconds": 11.965959372, "samples": 1 }, "passFailureRate": { "failedMeasuredPasses": 0, "totalMeasuredPasses": 16, "percent": 0 }, "attemptFailureRate": { "failedAttempts": 0, "totalAttempts": 16, "percent": 0 }, "roofline": { "modelWeightsGB": 17.987581215, "memoryBandwidthGBps": 760, "theoreticalMaxTokensPerSecond": 42.251372817498634, "utilization": 0.705799347019611 }, "diagnostics": [ { "id": "partial-cpu-offload", "status": "detected", "message": "59% of the model's resident bytes are on the host, not in VRAM", "evidence": { "source": "ollama.ps.size_vram", "granularity": "bytes", "residentBytes": 18814077917, "vramResidentBytes": 7806124031, "hostResidentBytes": 11007953886, "vramResidentFraction": 0.41490866921235364 } }, { "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": 30, "kvHeadCount": null, "attentionHeadCount": 16, "embeddingLength": 2816, "headDimension": 512, "resolvedElementType": null, "projectedBytes": null, "calculationAvailable": false, "missingInputs": [ "kvHeadCount", "resolvedKvCacheElementType" ] }, "missingInput": "resolvedKvCacheElementType" } }, { "id": "cpu-only-with-gpu", "status": "not-detected", "message": "41% of the model's resident bytes are in VRAM", "evidence": { "source": "ollama.ps.size_vram", "granularity": "bytes", "residentBytes": 18814077917, "vramResidentBytes": 7806124031, "hostResidentBytes": 11007953886, "vramResidentFraction": 0.41490866921235364, "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": 17987581215, "totalVramBytes": 10737418240 } } ] } }