{ "protocolVersion": "osai-bench/1.3", "clientVersion": "0.10.0", "scoringVersion": "osai-bench-derive/1.3", "createdAt": "2026-08-03T04:45:34.631Z", "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": 23781525417, "vramResidentBytes": 8269940652, "hostResidentBytes": 15511584765, "vramResidentFraction": 0.3477464337122929 }, "kvCacheMetadata": { "source": "ollama.show.model_info", "architecture": "qwen35moe", "blockCount": 40, "kvHeadCount": null, "attentionHeadCount": 16, "embeddingLength": 2048, "headDimension": 256, "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": "qwen3.6:35b-a3b", "digest": "07d35212591fc27746f0a317c975a6d68754fb38e9053d82e25f06057af28522", "family": "qwen35moe", "parameterSize": "36.0B", "quantization": "Q4_K_M", "weightsBytes": 23938333577, "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": 735051776, "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": "min_p 0\npresence_penalty 1.5\nrepeat_penalty 1\ntemperature 1\ntop_k 20\ntop_p 0.95", "unresolvedRuntimeDefaults": true } }, "rawMeasurements": { "workloads": { "w1": { "warmup": null, "measuredPasses": [ { "index": 1, "valid": true, "attempts": [ { "measurement": { "total_duration": 44294279296, "load_duration": 42776616922, "prompt_eval_count": 15, "prompt_eval_duration": 1502412000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 44297.15577799999 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 44294279296, "load_duration": 42776616922, "prompt_eval_count": 15, "prompt_eval_duration": 1502412000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 44297.15577799999 }, "validity": { "valid": true, "reasons": [] } } ], "failed": false }, "w2": { "warmup": { "total_duration": 46443114513, "load_duration": 41914378044, "prompt_eval_count": 49, "prompt_eval_duration": 745395000, "eval_count": 128, "eval_duration": 3781112000, "timeToFirstTokenMs": 42664.50010599999 }, "measuredPasses": [ { "index": 1, "valid": true, "attempts": [ { "measurement": { "total_duration": 4112466100, "load_duration": 224993212, "prompt_eval_count": 49, "prompt_eval_duration": 101469000, "eval_count": 128, "eval_duration": 3735095000, "timeToFirstTokenMs": 377.9849389999872 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 4112466100, "load_duration": 224993212, "prompt_eval_count": 49, "prompt_eval_duration": 101469000, "eval_count": 128, "eval_duration": 3735095000, "timeToFirstTokenMs": 377.9849389999872 }, "validity": { "valid": true, "reasons": [] } }, { "index": 2, "valid": true, "attempts": [ { "measurement": { "total_duration": 4077178998, "load_duration": 235745927, "prompt_eval_count": 49, "prompt_eval_duration": 103095000, "eval_count": 128, "eval_duration": 3736676000, "timeToFirstTokenMs": 341.1414689999947 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 4077178998, "load_duration": 235745927, "prompt_eval_count": 49, "prompt_eval_duration": 103095000, "eval_count": 128, "eval_duration": 3736676000, "timeToFirstTokenMs": 341.1414689999947 }, "validity": { "valid": true, "reasons": [] } }, { "index": 3, "valid": true, "attempts": [ { "measurement": { "total_duration": 4053126215, "load_duration": 237551224, "prompt_eval_count": 49, "prompt_eval_duration": 103342000, "eval_count": 128, "eval_duration": 3710509000, "timeToFirstTokenMs": 342.97518700000364 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 4053126215, "load_duration": 237551224, "prompt_eval_count": 49, "prompt_eval_duration": 103342000, "eval_count": 128, "eval_duration": 3710509000, "timeToFirstTokenMs": 342.97518700000364 }, "validity": { "valid": true, "reasons": [] } }, { "index": 4, "valid": true, "attempts": [ { "measurement": { "total_duration": 4115835945, "load_duration": 275403449, "prompt_eval_count": 49, "prompt_eval_duration": 103704000, "eval_count": 128, "eval_duration": 3734747000, "timeToFirstTokenMs": 381.47051800001645 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 4115835945, "load_duration": 275403449, "prompt_eval_count": 49, "prompt_eval_duration": 103704000, "eval_count": 128, "eval_duration": 3734747000, "timeToFirstTokenMs": 381.47051800001645 }, "validity": { "valid": true, "reasons": [] } }, { "index": 5, "valid": true, "attempts": [ { "measurement": { "total_duration": 4082590345, "load_duration": 237987905, "prompt_eval_count": 49, "prompt_eval_duration": 110346000, "eval_count": 128, "eval_duration": 3732292000, "timeToFirstTokenMs": 350.8226859999995 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 4082590345, "load_duration": 237987905, "prompt_eval_count": 49, "prompt_eval_duration": 110346000, "eval_count": 128, "eval_duration": 3732292000, "timeToFirstTokenMs": 350.8226859999995 }, "validity": { "valid": true, "reasons": [] } } ], "failed": false }, "w3": { "warmup": { "total_duration": 7063351237, "load_duration": 234862461, "prompt_eval_count": 2915, "prompt_eval_duration": 6810963000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 7063.938439999998 }, "measuredPasses": [ { "index": 1, "valid": true, "attempts": [ { "measurement": { "total_duration": 7172189638, "load_duration": 270408916, "prompt_eval_count": 2915, "prompt_eval_duration": 6795106000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 7172.669746 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 7172189638, "load_duration": 270408916, "prompt_eval_count": 2915, "prompt_eval_duration": 6795106000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 7172.669746 }, "validity": { "valid": true, "reasons": [] } }, { "index": 2, "valid": true, "attempts": [ { "measurement": { "total_duration": 7135868546, "load_duration": 258406035, "prompt_eval_count": 2915, "prompt_eval_duration": 6757086000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 7136.175713000004 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 7135868546, "load_duration": 258406035, "prompt_eval_count": 2915, "prompt_eval_duration": 6757086000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 7136.175713000004 }, "validity": { "valid": true, "reasons": [] } }, { "index": 3, "valid": true, "attempts": [ { "measurement": { "total_duration": 7146082629, "load_duration": 276267362, "prompt_eval_count": 2915, "prompt_eval_duration": 6760038000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 7146.311172999995 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 7146082629, "load_duration": 276267362, "prompt_eval_count": 2915, "prompt_eval_duration": 6760038000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 7146.311172999995 }, "validity": { "valid": true, "reasons": [] } }, { "index": 4, "valid": true, "attempts": [ { "measurement": { "total_duration": 7104751647, "load_duration": 235501491, "prompt_eval_count": 2915, "prompt_eval_duration": 6746384000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 7105.2407759999915 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 7104751647, "load_duration": 235501491, "prompt_eval_count": 2915, "prompt_eval_duration": 6746384000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 7105.2407759999915 }, "validity": { "valid": true, "reasons": [] } }, { "index": 5, "valid": true, "attempts": [ { "measurement": { "total_duration": 7108949162, "load_duration": 237822568, "prompt_eval_count": 2915, "prompt_eval_duration": 6755209000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 7109.219027000014 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 7108949162, "load_duration": 237822568, "prompt_eval_count": 2915, "prompt_eval_duration": 6755209000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 7109.219027000014 }, "validity": { "valid": true, "reasons": [] } } ], "failed": false }, "w4": { "warmup": { "total_duration": 15466137942, "load_duration": 241461164, "prompt_eval_count": 49, "prompt_eval_duration": 100380000, "eval_count": 512, "eval_duration": 15015028000, "timeToFirstTokenMs": 451.56861899999785 }, "measuredPasses": [ { "index": 1, "valid": true, "attempts": [ { "measurement": { "total_duration": 15493636755, "load_duration": 246260528, "prompt_eval_count": 49, "prompt_eval_duration": 102572000, "eval_count": 512, "eval_duration": 15070653000, "timeToFirstTokenMs": 423.5229319999926 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 15493636755, "load_duration": 246260528, "prompt_eval_count": 49, "prompt_eval_duration": 102572000, "eval_count": 512, "eval_duration": 15070653000, "timeToFirstTokenMs": 423.5229319999926 }, "validity": { "valid": true, "reasons": [] } }, { "index": 2, "valid": true, "attempts": [ { "measurement": { "total_duration": 15299997195, "load_duration": 247063100, "prompt_eval_count": 49, "prompt_eval_duration": 105651000, "eval_count": 512, "eval_duration": 14945404000, "timeToFirstTokenMs": 354.90692000000854 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 15299997195, "load_duration": 247063100, "prompt_eval_count": 49, "prompt_eval_duration": 105651000, "eval_count": 512, "eval_duration": 14945404000, "timeToFirstTokenMs": 354.90692000000854 }, "validity": { "valid": true, "reasons": [] } }, { "index": 3, "valid": true, "attempts": [ { "measurement": { "total_duration": 15286562110, "load_duration": 238352727, "prompt_eval_count": 49, "prompt_eval_duration": 103165000, "eval_count": 512, "eval_duration": 14943163000, "timeToFirstTokenMs": 343.7073960000125 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 15286562110, "load_duration": 238352727, "prompt_eval_count": 49, "prompt_eval_duration": 103165000, "eval_count": 512, "eval_duration": 14943163000, "timeToFirstTokenMs": 343.7073960000125 }, "validity": { "valid": true, "reasons": [] } }, { "index": 4, "valid": true, "attempts": [ { "measurement": { "total_duration": 15292373878, "load_duration": 238192402, "prompt_eval_count": 49, "prompt_eval_duration": 103146000, "eval_count": 512, "eval_duration": 14949135000, "timeToFirstTokenMs": 343.7275579999841 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 15292373878, "load_duration": 238192402, "prompt_eval_count": 49, "prompt_eval_duration": 103146000, "eval_count": 512, "eval_duration": 14949135000, "timeToFirstTokenMs": 343.7275579999841 }, "validity": { "valid": true, "reasons": [] } }, { "index": 5, "valid": true, "attempts": [ { "measurement": { "total_duration": 15298339934, "load_duration": 233233943, "prompt_eval_count": 49, "prompt_eval_duration": 104106000, "eval_count": 512, "eval_duration": 14959140000, "timeToFirstTokenMs": 339.6978019999806 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 15298339934, "load_duration": 233233943, "prompt_eval_count": 49, "prompt_eval_duration": 104106000, "eval_count": 512, "eval_duration": 14959140000, "timeToFirstTokenMs": 339.6978019999806 }, "validity": { "valid": true, "reasons": [] } } ], "failed": false } } }, "derived": { "generationTokensPerSecond": { "median": 34.249473297284425, "coefficientOfVariation": 0.003632810379113527, "samples": 5 }, "prefillTokensPerSecond": { "median": 431.3989787905615, "coefficientOfVariation": 0.002768926339272169, "samples": 5 }, "timeToFirstTokenMs": { "median": 350.8226859999995, "coefficientOfVariation": 0.054100816779501036, "samples": 5 }, "coldLoad": { "seconds": 42.776616922, "samples": 1 }, "passFailureRate": { "failedMeasuredPasses": 0, "totalMeasuredPasses": 16, "percent": 0 }, "attemptFailureRate": { "failedAttempts": 0, "totalAttempts": 16, "percent": 0 }, "roofline": { "modelWeightsGB": 23.938333577, "memoryBandwidthGBps": 760, "theoreticalMaxTokensPerSecond": 31.7482416875588, "utilization": 1.0787833113512484 }, "diagnostics": [ { "id": "partial-cpu-offload", "status": "detected", "message": "65% of the model's resident bytes are on the host, not in VRAM", "evidence": { "source": "ollama.ps.size_vram", "granularity": "bytes", "residentBytes": 23781525417, "vramResidentBytes": 8269940652, "hostResidentBytes": 15511584765, "vramResidentFraction": 0.3477464337122929 } }, { "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": "qwen35moe", "blockCount": 40, "kvHeadCount": null, "attentionHeadCount": 16, "embeddingLength": 2048, "headDimension": 256, "resolvedElementType": null, "projectedBytes": null, "calculationAvailable": false, "missingInputs": [ "kvHeadCount", "resolvedKvCacheElementType" ] }, "missingInput": "resolvedKvCacheElementType" } }, { "id": "cpu-only-with-gpu", "status": "not-detected", "message": "35% of the model's resident bytes are in VRAM", "evidence": { "source": "ollama.ps.size_vram", "granularity": "bytes", "residentBytes": 23781525417, "vramResidentBytes": 8269940652, "hostResidentBytes": 15511584765, "vramResidentFraction": 0.3477464337122929, "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": 23938333577, "totalVramBytes": 10737418240 } } ] } }