{"name":"Granite-4.1 8B","arena":{"id":"granite-4-1-8b-bf16","name":"Granite-4.1 8B","vram":16,"params":8,"vendor":"ibm","quality":{"coding":{"bench":"HumanEval","value":85.37},"science":{"bench":"GPQA","value":41.96},"knowledge":{"bench":"MMLU-Pro","value":55.99}},"contextK":128},"notes":"Model-ID: ibm-granite/granite-4.1-8b. Container: vllm/vllm-openai:v0.26.0-aarch64-cu129-ubuntu2404. llama-benchy 0.4.0. Bronrun: 2026-08-09T06:26:37+02:00. KV-cache fp8; max_model_len 131072; gpu_memory_utilization 0,9. Prefix caching uit; async scheduling aan. Extra serverflags: --max-num-batched-tokens 8192. 11 van de 11 tests hebben een meting die volgens de huidige Arena-regels meetelt. Bij deze historische run is geen sanity-check vastgelegd. Dat bewijs ontbreekt, ook wanneer de snelheidsmeting volgens de huidige regels meetelt.","order":100,"engine":"vLLM v0.26.0","verdict":"Gemeten op 2026-08-09 met vLLM v0.26.0. In de chattest met tien gelijktijdige verzoeken haalt dit profiel 11,99 tokens/s per gebruiker, met gemiddeld 1,89 seconden tot het eerste token. Bij 25k context stijgt die gemiddelde wachttijd naar 39,32 seconden. Bij deze historische run is geen sanity-check vastgelegd. Dat bewijs ontbreekt, ook wanneer de snelheidsmeting volgens de huidige regels meetelt.","hardware":"DGX Spark, NVIDIA GB10, 128 GB unified memory","modelUrl":"https://huggingface.co/ibm-granite/granite-4.1-8b","provider":"ibm-granite","precision":"BF16","parameters":"8B","provenance":{"modelId":"ibm-granite/granite-4.1-8b","runPath":"results/granite-4.1/granite-4.1-8b/bf16","validity":{"tests":{"01-chat":"complete","02-rag-8k":"complete","08-sharegpt":"complete","09-reasoning":"complete","04-multi-turn":"complete","11-rate-sweep":"complete","03-long-output":"complete","05-big-context":"complete","10-monday-peak":"complete","07-office-baseline":"complete","06-concurrency-stress":"complete"},"sanity":"not_recorded"},"sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/granite-4.1/granite-4.1-8b/bf16/","servedName":"granite-4-1-8b-bf16","generatedAt":"2026-08-09T06:26:37+02:00","runRevision":"a948cec23b054c62fef595bad3cdb284066922ac","vllmVersion":"0.26.0","suiteVersion":"2026-08","repositoryUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks","contractVersion":1,"contractRevision":"a948cec23b054c62fef595bad3cdb284066922ac","commandProvenance":"reconstructed_from_raw_artifacts","llamaBenchyVersion":"0.4.0"},"architecture":"Dense","measuredContextK":128,"benchmarkResults":[{"id":"granite-4-1-8b-bf16-chat-1k","date":"2026-08-09","runs":3,"metrics":{"ttfrSec":{"mean":1.887,"stddev":0.689},"decodeTokensPerSecondTotal":{"mean":60.09,"stddev":7},"prefillTokensPerSecondTotal":{"mean":3576.91,"stddev":216.58},"decodeTokensPerSecondPerUser":{"mean":11.99,"stddev":0.5}},"sourceRun":"Test 01","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/granite-4.1/granite-4.1-8b/bf16/","workloadId":"chat-1k","sourceLabel":"ibm-granite/granite-4.1-8b bf16 op de DGX Spark","sourceStatus":"complete","sourceTestId":"01-chat","prefixCaching":false},{"id":"granite-4-1-8b-bf16-rag-8k","date":"2026-08-09","runs":3,"metrics":{"ttfrSec":{"mean":10.825,"stddev":5.665},"decodeTokensPerSecondTotal":{"mean":69.03,"stddev":3.7},"prefillTokensPerSecondTotal":{"mean":3905.43,"stddev":2.59},"decodeTokensPerSecondPerUser":{"mean":8.62,"stddev":1.45}},"sourceRun":"Test 02","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/granite-4.1/granite-4.1-8b/bf16/","workloadId":"rag-8k","sourceLabel":"ibm-granite/granite-4.1-8b bf16 op de DGX Spark","sourceStatus":"complete","sourceTestId":"02-rag-8k","prefixCaching":false},{"id":"granite-4-1-8b-bf16-long-output-agents","date":"2026-08-09","runs":3,"metrics":{"ttfrSec":{"mean":0.71,"stddev":0.004},"decodeTokensPerSecondTotal":{"mean":75.84,"stddev":7.24},"prefillTokensPerSecondTotal":{"mean":3263.85,"stddev":23.36},"decodeTokensPerSecondPerUser":{"mean":13.03,"stddev":0.03}},"sourceRun":"Test 03","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/granite-4.1/granite-4.1-8b/bf16/","workloadId":"long-output-agents","sourceLabel":"ibm-granite/granite-4.1-8b bf16 op de DGX Spark","sourceStatus":"complete","sourceTestId":"03-long-output","prefixCaching":false},{"id":"granite-4-1-8b-bf16-multi-turn-office","date":"2026-08-09","runs":3,"metrics":{"ttfrSec":{"mean":3.158,"stddev":1.245},"decodeTokensPerSecondTotal":{"mean":99.94,"stddev":3.96},"prefillTokensPerSecondTotal":{"mean":4236.98,"stddev":81.26},"decodeTokensPerSecondPerUser":{"mean":11.86,"stddev":0.34}},"sourceRun":"Test 04","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/granite-4.1/granite-4.1-8b/bf16/","workloadId":"multi-turn-office","sourceLabel":"ibm-granite/granite-4.1-8b bf16 op de DGX Spark","sourceStatus":"complete","sourceTestId":"04-multi-turn","prefixCaching":false},{"id":"granite-4-1-8b-bf16-large-context-25k","date":"2026-08-09","runs":3,"metrics":{"ttfrSec":{"mean":39.324,"stddev":20.012},"decodeTokensPerSecondTotal":{"mean":24.75,"stddev":0.46},"prefillTokensPerSecondTotal":{"mean":3154,"stddev":5.77},"decodeTokensPerSecondPerUser":{"mean":4.11,"stddev":1.38}},"sourceRun":"Test 05","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/granite-4.1/granite-4.1-8b/bf16/","workloadId":"large-context-25k","sourceLabel":"ibm-granite/granite-4.1-8b bf16 op de DGX Spark","sourceStatus":"complete","sourceTestId":"05-big-context","prefixCaching":false},{"id":"granite-4-1-8b-bf16-concurrency-stress-25k","date":"2026-08-09","runs":3,"metrics":{"ttfrSec":{"mean":74.995,"stddev":41.05},"decodeTokensPerSecondTotal":{"mean":25.97,"stddev":0.25},"prefillTokensPerSecondTotal":{"mean":3102.83,"stddev":1.5},"decodeTokensPerSecondPerUser":{"mean":2.31,"stddev":0.86}},"sourceRun":"Test 06","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/granite-4.1/granite-4.1-8b/bf16/","workloadId":"concurrency-stress-25k","sourceLabel":"ibm-granite/granite-4.1-8b bf16 op de DGX Spark","sourceStatus":"complete","sourceTestId":"06-concurrency-stress","prefixCaching":false}],"openLoopResults":[{"id":"granite-4-1-8b-bf16-office-random-4k","date":"2026-08-09","metrics":{"e2eMs":{"p50":89746.87,"p90":155638.97,"p95":169377.03,"p99":180850.05,"mean":88860.68},"tpotMs":{"p50":183.63,"p90":205.57,"p95":213.12,"p99":227.79,"mean":182.14},"ttftMs":{"p50":1751.45,"p90":3445.24,"p95":3811.88,"p99":4852.04,"mean":1956.83}},"sourceRun":"Test 07","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/granite-4.1/granite-4.1-8b/bf16/","scenarioId":"office-random-4k","achievedRps":0.266,"sourceStatus":"complete","sourceTestId":"07-office-baseline","configuredRps":0.3,"prefixCaching":false,"totalRequests":200,"successfulRequests":200,"totalTokenThroughput":1213.47,"peakConcurrentRequests":39},{"id":"granite-4-1-8b-bf16-sharegpt-replay","date":"2026-08-09","metrics":{"e2eMs":{"p50":10057.66,"p90":37013.76,"p95":46045.56,"p99":59695.31,"mean":15324.67},"tpotMs":{"p50":77.45,"p90":79.33,"p95":80.82,"p99":93.85,"mean":78.05},"ttftMs":{"p50":254.56,"p90":342.8,"p95":384.71,"p99":435.94,"mean":266.48}},"sourceRun":"Test 08","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/granite-4.1/granite-4.1-8b/bf16/","scenarioId":"sharegpt-replay","achievedRps":0.291,"sourceStatus":"complete","sourceTestId":"08-sharegpt","configuredRps":0.3,"prefixCaching":false,"totalRequests":250,"successfulRequests":250,"totalTokenThroughput":121.43,"peakConcurrentRequests":13},{"id":"granite-4-1-8b-bf16-reasoning","date":"2026-08-09","metrics":{"e2eMs":{"p50":518990.96,"p90":851193.17,"p95":915004.24,"p99":938335.79,"mean":544873.26},"tpotMs":{"p50":131.03,"p90":138.3,"p95":139.14,"p99":140.1,"mean":129.11},"ttftMs":{"p50":523.19,"p90":734.61,"p95":763.28,"p99":872.5,"mean":543.02}},"sourceRun":"Test 09","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/granite-4.1/granite-4.1-8b/bf16/","scenarioId":"reasoning","achievedRps":0.045,"sourceStatus":"complete","sourceTestId":"09-reasoning","configuredRps":0.2,"prefixCaching":false,"totalRequests":50,"successfulRequests":50,"totalTokenThroughput":240.52,"peakConcurrentRequests":46},{"id":"granite-4-1-8b-bf16-monday-peak-random-4k","date":"2026-08-09","metrics":{"e2eMs":{"p50":86317.72,"p90":158057.75,"p95":164563.02,"p99":173935.34,"mean":88793.07},"tpotMs":{"p50":175.65,"p90":188.4,"p95":199.16,"p99":233.22,"mean":175.86},"ttftMs":{"p50":1489.86,"p90":2487.83,"p95":4944.12,"p99":12800.91,"mean":1930.22}},"sourceRun":"Test 10","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/granite-4.1/granite-4.1-8b/bf16/","scenarioId":"monday-peak-random-4k","achievedRps":0.271,"sourceStatus":"complete","sourceTestId":"10-monday-peak","configuredRps":1.5,"prefixCaching":false,"totalRequests":300,"successfulRequests":300,"totalTokenThroughput":1218.28,"peakConcurrentRequests":28}],"rateSweep":{"date":"2026-08-09","capacity":{"slo2000":null,"slo5000":{"outputTps":115.1,"ttftP95Ms":4638.3,"achievedRps":0.234,"configuredRps":0.3,"peakConcurrent":32},"slo10000":{"outputTps":191,"ttftP95Ms":8463.8,"achievedRps":0.391,"configuredRps":0.7,"peakConcurrent":153}},"sourceRun":"Test 11","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/granite-4.1/granite-4.1-8b/bf16/","abortedAtRps":null,"queueKneeRps":null,"sourceStatus":"complete","sourceTestId":"11-rate-sweep","totalRequests":850,"successfulRequests":850}}