{ "protocolVersion": "osai-bench/1.3", "clientVersion": "0.10.0", "scoringVersion": "osai-bench-derive/1.3", "createdAt": "2026-08-03T05:09:01.668Z", "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": 17399923338, "vramResidentBytes": 8050148636, "hostResidentBytes": 9349774702, "vramResidentFraction": 0.4626542588506202 }, "kvCacheMetadata": { "source": "ollama.show.model_info", "architecture": "qwen35", "blockCount": 64, "kvHeadCount": null, "attentionHeadCount": 24, "embeddingLength": 5120, "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": "qwen3.6:27b", "digest": "a50eda8ed977ab48a12431878896b27ffd5cef552c17af3317d9623b939a7f1e", "family": "qwen35", "parameterSize": "27.8B", "quantization": "Q4_K_M", "weightsBytes": 17420432739, "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": 936378368, "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": "presence_penalty 1.5\nrepeat_penalty 1\ntemperature 1\ntop_k 20\ntop_p 0.95\nmin_p 0", "unresolvedRuntimeDefaults": true } }, "rawMeasurements": { "workloads": { "w1": { "warmup": null, "measuredPasses": [ { "index": 1, "valid": true, "attempts": [ { "measurement": { "total_duration": 33295501222, "load_duration": 32151884053, "prompt_eval_count": 15, "prompt_eval_duration": 1141676000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 33296.272591 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 33295501222, "load_duration": 32151884053, "prompt_eval_count": 15, "prompt_eval_duration": 1141676000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 33296.272591 }, "validity": { "valid": true, "reasons": [] } } ], "failed": false }, "w2": { "warmup": { "total_duration": 75418177482, "load_duration": 32882043610, "prompt_eval_count": 49, "prompt_eval_duration": 1362245000, "eval_count": 128, "eval_duration": 41171939000, "timeToFirstTokenMs": 34247.73022899999 }, "measuredPasses": [ { "index": 1, "valid": true, "attempts": [ { "measurement": { "total_duration": 41985474834, "load_duration": 246007482, "prompt_eval_count": 49, "prompt_eval_duration": 460654000, "eval_count": 128, "eval_duration": 41159111000, "timeToFirstTokenMs": 827.1597789999796 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 41985474834, "load_duration": 246007482, "prompt_eval_count": 49, "prompt_eval_duration": 460654000, "eval_count": 128, "eval_duration": 41159111000, "timeToFirstTokenMs": 827.1597789999796 }, "validity": { "valid": true, "reasons": [] } }, { "index": 2, "valid": true, "attempts": [ { "measurement": { "total_duration": 41852567019, "load_duration": 235757544, "prompt_eval_count": 49, "prompt_eval_duration": 462689000, "eval_count": 128, "eval_duration": 41152225000, "timeToFirstTokenMs": 700.8671050000121 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 41852567019, "load_duration": 235757544, "prompt_eval_count": 49, "prompt_eval_duration": 462689000, "eval_count": 128, "eval_duration": 41152225000, "timeToFirstTokenMs": 700.8671050000121 }, "validity": { "valid": true, "reasons": [] } }, { "index": 3, "valid": true, "attempts": [ { "measurement": { "total_duration": 41841033282, "load_duration": 234661281, "prompt_eval_count": 49, "prompt_eval_duration": 462414000, "eval_count": 128, "eval_duration": 41141999000, "timeToFirstTokenMs": 699.2819519999903 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 41841033282, "load_duration": 234661281, "prompt_eval_count": 49, "prompt_eval_duration": 462414000, "eval_count": 128, "eval_duration": 41141999000, "timeToFirstTokenMs": 699.2819519999903 }, "validity": { "valid": true, "reasons": [] } }, { "index": 4, "valid": true, "attempts": [ { "measurement": { "total_duration": 41865617929, "load_duration": 247908500, "prompt_eval_count": 49, "prompt_eval_duration": 464121000, "eval_count": 128, "eval_duration": 41151284000, "timeToFirstTokenMs": 714.5137899999972 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 41865617929, "load_duration": 247908500, "prompt_eval_count": 49, "prompt_eval_duration": 464121000, "eval_count": 128, "eval_duration": 41151284000, "timeToFirstTokenMs": 714.5137899999972 }, "validity": { "valid": true, "reasons": [] } }, { "index": 5, "valid": true, "attempts": [ { "measurement": { "total_duration": 41839408866, "load_duration": 236597647, "prompt_eval_count": 49, "prompt_eval_duration": 461131000, "eval_count": 128, "eval_duration": 41139539000, "timeToFirstTokenMs": 700.1652079999913 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 41839408866, "load_duration": 236597647, "prompt_eval_count": 49, "prompt_eval_duration": 461131000, "eval_count": 128, "eval_duration": 41139539000, "timeToFirstTokenMs": 700.1652079999913 }, "validity": { "valid": true, "reasons": [] } } ], "failed": false }, "w3": { "warmup": { "total_duration": 8411680382, "load_duration": 240505081, "prompt_eval_count": 2915, "prompt_eval_duration": 8144472000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 8412.121668000007 }, "measuredPasses": [ { "index": 1, "valid": true, "attempts": [ { "measurement": { "total_duration": 8656119480, "load_duration": 274130210, "prompt_eval_count": 2915, "prompt_eval_duration": 8133744000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 8656.534656000033 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 8656119480, "load_duration": 274130210, "prompt_eval_count": 2915, "prompt_eval_duration": 8133744000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 8656.534656000033 }, "validity": { "valid": true, "reasons": [] } }, { "index": 2, "valid": true, "attempts": [ { "measurement": { "total_duration": 8670589988, "load_duration": 280074558, "prompt_eval_count": 2915, "prompt_eval_duration": 8136542000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 8670.847451000009 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 8670589988, "load_duration": 280074558, "prompt_eval_count": 2915, "prompt_eval_duration": 8136542000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 8670.847451000009 }, "validity": { "valid": true, "reasons": [] } }, { "index": 3, "valid": true, "attempts": [ { "measurement": { "total_duration": 8629931756, "load_duration": 237183318, "prompt_eval_count": 2915, "prompt_eval_duration": 8137241000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 8630.411194999993 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 8629931756, "load_duration": 237183318, "prompt_eval_count": 2915, "prompt_eval_duration": 8137241000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 8630.411194999993 }, "validity": { "valid": true, "reasons": [] } }, { "index": 4, "valid": true, "attempts": [ { "measurement": { "total_duration": 8683259068, "load_duration": 271128864, "prompt_eval_count": 2915, "prompt_eval_duration": 8141381000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 8683.761193000013 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 8683259068, "load_duration": 271128864, "prompt_eval_count": 2915, "prompt_eval_duration": 8141381000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 8683.761193000013 }, "validity": { "valid": true, "reasons": [] } }, { "index": 5, "valid": true, "attempts": [ { "measurement": { "total_duration": 8627354344, "load_duration": 234713862, "prompt_eval_count": 2915, "prompt_eval_duration": 8146125000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 8627.738020999997 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 8627354344, "load_duration": 234713862, "prompt_eval_count": 2915, "prompt_eval_duration": 8146125000, "eval_count": 1, "eval_duration": 1000, "timeToFirstTokenMs": 8627.738020999997 }, "validity": { "valid": true, "reasons": [] } } ], "failed": false }, "w4": { "warmup": { "total_duration": 167162349821, "load_duration": 242824278, "prompt_eval_count": 49, "prompt_eval_duration": 454834000, "eval_count": 512, "eval_duration": 166202586000, "timeToFirstTokenMs": 960.3160420000204 }, "measuredPasses": [ { "index": 1, "valid": true, "attempts": [ { "measurement": { "total_duration": 167141752743, "load_duration": 264610119, "prompt_eval_count": 49, "prompt_eval_duration": 458549000, "eval_count": 512, "eval_duration": 166246635000, "timeToFirstTokenMs": 895.6671939999796 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 167141752743, "load_duration": 264610119, "prompt_eval_count": 49, "prompt_eval_duration": 458549000, "eval_count": 512, "eval_duration": 166246635000, "timeToFirstTokenMs": 895.6671939999796 }, "validity": { "valid": true, "reasons": [] } }, { "index": 2, "valid": true, "attempts": [ { "measurement": { "total_duration": 166937328401, "load_duration": 253936981, "prompt_eval_count": 49, "prompt_eval_duration": 461974000, "eval_count": 512, "eval_duration": 166219593000, "timeToFirstTokenMs": 718.2526539999526 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 166937328401, "load_duration": 253936981, "prompt_eval_count": 49, "prompt_eval_duration": 461974000, "eval_count": 512, "eval_duration": 166219593000, "timeToFirstTokenMs": 718.2526539999526 }, "validity": { "valid": true, "reasons": [] } }, { "index": 3, "valid": true, "attempts": [ { "measurement": { "total_duration": 166878341567, "load_duration": 262358808, "prompt_eval_count": 49, "prompt_eval_duration": 461958000, "eval_count": 512, "eval_duration": 166151845000, "timeToFirstTokenMs": 726.95597000001 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 166878341567, "load_duration": 262358808, "prompt_eval_count": 49, "prompt_eval_duration": 461958000, "eval_count": 512, "eval_duration": 166151845000, "timeToFirstTokenMs": 726.95597000001 }, "validity": { "valid": true, "reasons": [] } }, { "index": 4, "valid": true, "attempts": [ { "measurement": { "total_duration": 166868284363, "load_duration": 255439136, "prompt_eval_count": 49, "prompt_eval_duration": 463068000, "eval_count": 512, "eval_duration": 166148157000, "timeToFirstTokenMs": 720.6229280000553 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 166868284363, "load_duration": 255439136, "prompt_eval_count": 49, "prompt_eval_duration": 463068000, "eval_count": 512, "eval_duration": 166148157000, "timeToFirstTokenMs": 720.6229280000553 }, "validity": { "valid": true, "reasons": [] } }, { "index": 5, "valid": true, "attempts": [ { "measurement": { "total_duration": 166937118592, "load_duration": 244442435, "prompt_eval_count": 49, "prompt_eval_duration": 460435000, "eval_count": 512, "eval_duration": 166230366000, "timeToFirstTokenMs": 707.2284119999968 }, "validity": { "valid": true, "reasons": [] } } ], "measurement": { "total_duration": 166937118592, "load_duration": 244442435, "prompt_eval_count": 49, "prompt_eval_duration": 460435000, "eval_count": 512, "eval_duration": 166230366000, "timeToFirstTokenMs": 707.2284119999968 }, "validity": { "valid": true, "reasons": [] } } ], "failed": false } } }, "derived": { "generationTokensPerSecond": { "median": 3.0802626258385795, "coefficientOfVariation": 0.0002771396643681258, "samples": 5 }, "prefillTokensPerSecond": { "median": 358.22952767406053, "coefficientOfVariation": 0.0005928956336707891, "samples": 5 }, "timeToFirstTokenMs": { "median": 700.8671050000121, "coefficientOfVariation": 0.07628255705679458, "samples": 5 }, "coldLoad": { "seconds": 32.151884053, "samples": 1 }, "passFailureRate": { "failedMeasuredPasses": 0, "totalMeasuredPasses": 16, "percent": 0 }, "attemptFailureRate": { "failedAttempts": 0, "totalAttempts": 16, "percent": 0 }, "roofline": { "modelWeightsGB": 17.420432739, "memoryBandwidthGBps": 760, "theoreticalMaxTokensPerSecond": 43.62693001871014, "utilization": 0.07060461564720591 }, "diagnostics": [ { "id": "partial-cpu-offload", "status": "detected", "message": "54% of the model's resident bytes are on the host, not in VRAM", "evidence": { "source": "ollama.ps.size_vram", "granularity": "bytes", "residentBytes": 17399923338, "vramResidentBytes": 8050148636, "hostResidentBytes": 9349774702, "vramResidentFraction": 0.4626542588506202 } }, { "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": "qwen35", "blockCount": 64, "kvHeadCount": null, "attentionHeadCount": 24, "embeddingLength": 5120, "headDimension": 256, "resolvedElementType": null, "projectedBytes": null, "calculationAvailable": false, "missingInputs": [ "kvHeadCount", "resolvedKvCacheElementType" ] }, "missingInput": "resolvedKvCacheElementType" } }, { "id": "cpu-only-with-gpu", "status": "not-detected", "message": "46% of the model's resident bytes are in VRAM", "evidence": { "source": "ollama.ps.size_vram", "granularity": "bytes", "residentBytes": 17399923338, "vramResidentBytes": 8050148636, "hostResidentBytes": 9349774702, "vramResidentFraction": 0.4626542588506202, "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": 17420432739, "totalVramBytes": 10737418240 } } ] } }