{"name":"Nemotron-3 Nano Omni 30B-A3B","arena":{"id":"nemotron-3-nano-omni-30b-a3b-reasoning-fp8","name":"Nemotron-3 Nano Omni 30B-A3B","vram":33,"params":30,"vendor":"nvidia","quality":{"coding":null,"science":null,"knowledge":null},"contextK":256,"baseModelId":"nemotron-3-nano-omni-30b-a3b-reasoning"},"notes":"Model-ID: nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-FP8. Container: vllm/vllm-openai:v0.26.0-aarch64-cu129-ubuntu2404. llama-benchy 0.4.0. Bronrun: 2026-08-06T03:41:08+02:00. KV-cache fp8; max_model_len 131072; gpu_memory_utilization 0,9. Prefix caching uit; async scheduling uit. Extra serverflags: --trust-remote-code --max-num-seqs 256 --max-num-batched-tokens 8192 --video-pruning-rate 0.5 --reasoning-parser nemotron_v3 --enable-auto-tool-choice --tool-call-parser qwen3_coder. 9 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. Geen brondata voor: 11-rate-sweep.","order":240,"engine":"vLLM v0.26.0","verdict":"Gemeten op 2026-08-06 met vLLM v0.26.0. In de chattest met tien gelijktijdige verzoeken haalt dit profiel 16,85 tokens/s per gebruiker, met gemiddeld 1,42 seconden tot het eerste token. Bij 25k context stijgt die gemiddelde wachttijd naar 14,32 seconden. Bij deze historische run is geen sanity-check vastgelegd. Dat bewijs ontbreekt, ook wanneer de snelheidsmeting volgens de huidige regels meetelt. De piekmeting is uitgesloten omdat minder dan 99% van de verzoeken slaagt. Voor de rate sweep ontbreekt brondata; daaruit volgt geen conclusie over lage capaciteit.","hardware":"DGX Spark, NVIDIA GB10, 128 GB unified memory","modelUrl":"https://huggingface.co/nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-FP8","provider":"nvidia","precision":"FP8","parameters":"30B-A3B","provenance":{"modelId":"nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-FP8","runPath":"results/nemotron-3/nemotron-3-nano-omni-30b-a3b-reasoning/fp8","validity":{"tests":{"01-chat":"complete","02-rag-8k":"complete","08-sharegpt":"complete","09-reasoning":"complete","04-multi-turn":"complete","11-rate-sweep":"no_data","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/nemotron-3/nemotron-3-nano-omni-30b-a3b-reasoning/fp8/","servedName":"nemotron-3-nano-omni-30b-fp8","generatedAt":"2026-08-06T03:41:08+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":"MoE","measuredContextK":128,"benchmarkResults":[{"id":"nemotron-3-nano-omni-30b-a3b-reasoning-fp8-chat-1k","date":"2026-08-06","runs":3,"metrics":{"ttfrSec":{"mean":1.416,"stddev":0.243},"decodeTokensPerSecondTotal":{"mean":145.09,"stddev":6.52},"prefillTokensPerSecondTotal":{"mean":5866.47,"stddev":93.08},"decodeTokensPerSecondPerUser":{"mean":16.85,"stddev":1.34}},"sourceRun":"Test 01","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/nemotron-3/nemotron-3-nano-omni-30b-a3b-reasoning/fp8/","workloadId":"chat-1k","sourceLabel":"nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-FP8 fp8 op de DGX Spark","sourceStatus":"complete","sourceTestId":"01-chat","prefixCaching":false},{"id":"nemotron-3-nano-omni-30b-a3b-reasoning-fp8-rag-8k","date":"2026-08-06","runs":3,"metrics":{"ttfrSec":{"mean":5.156,"stddev":2.379},"decodeTokensPerSecondTotal":{"mean":117.1,"stddev":0.77},"prefillTokensPerSecondTotal":{"mean":8683.04,"stddev":176.32},"decodeTokensPerSecondPerUser":{"mean":13.44,"stddev":0.87}},"sourceRun":"Test 02","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/nemotron-3/nemotron-3-nano-omni-30b-a3b-reasoning/fp8/","workloadId":"rag-8k","sourceLabel":"nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-FP8 fp8 op de DGX Spark","sourceStatus":"complete","sourceTestId":"02-rag-8k","prefixCaching":false},{"id":"nemotron-3-nano-omni-30b-a3b-reasoning-fp8-long-output-agents","date":"2026-08-06","runs":3,"metrics":{"ttfrSec":{"mean":0.863,"stddev":0.074},"decodeTokensPerSecondTotal":{"mean":120.57,"stddev":4.12},"prefillTokensPerSecondTotal":{"mean":2657.5,"stddev":50.14},"decodeTokensPerSecondPerUser":{"mean":18.75,"stddev":3.67}},"sourceRun":"Test 03","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/nemotron-3/nemotron-3-nano-omni-30b-a3b-reasoning/fp8/","workloadId":"long-output-agents","sourceLabel":"nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-FP8 fp8 op de DGX Spark","sourceStatus":"complete","sourceTestId":"03-long-output","prefixCaching":false},{"id":"nemotron-3-nano-omni-30b-a3b-reasoning-fp8-multi-turn-office","date":"2026-08-06","runs":3,"metrics":{"ttfrSec":{"mean":1.998,"stddev":0.633},"decodeTokensPerSecondTotal":{"mean":137.26,"stddev":1.26},"prefillTokensPerSecondTotal":{"mean":7044.62,"stddev":498.96},"decodeTokensPerSecondPerUser":{"mean":14.81,"stddev":0.64}},"sourceRun":"Test 04","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/nemotron-3/nemotron-3-nano-omni-30b-a3b-reasoning/fp8/","workloadId":"multi-turn-office","sourceLabel":"nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-FP8 fp8 op de DGX Spark","sourceStatus":"complete","sourceTestId":"04-multi-turn","prefixCaching":false},{"id":"nemotron-3-nano-omni-30b-a3b-reasoning-fp8-large-context-25k","date":"2026-08-06","runs":3,"metrics":{"ttfrSec":{"mean":14.318,"stddev":7.211},"decodeTokensPerSecondTotal":{"mean":62.63,"stddev":0.8},"prefillTokensPerSecondTotal":{"mean":8819.15,"stddev":13.3},"decodeTokensPerSecondPerUser":{"mean":9.69,"stddev":2.75}},"sourceRun":"Test 05","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/nemotron-3/nemotron-3-nano-omni-30b-a3b-reasoning/fp8/","workloadId":"large-context-25k","sourceLabel":"nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-FP8 fp8 op de DGX Spark","sourceStatus":"complete","sourceTestId":"05-big-context","prefixCaching":false},{"id":"nemotron-3-nano-omni-30b-a3b-reasoning-fp8-concurrency-stress-25k","date":"2026-08-06","runs":3,"metrics":{"ttfrSec":{"mean":26.828,"stddev":14.47},"decodeTokensPerSecondTotal":{"mean":70.65,"stddev":1.46},"prefillTokensPerSecondTotal":{"mean":8884.14,"stddev":12.59},"decodeTokensPerSecondPerUser":{"mean":6.28,"stddev":2.21}},"sourceRun":"Test 06","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/nemotron-3/nemotron-3-nano-omni-30b-a3b-reasoning/fp8/","workloadId":"concurrency-stress-25k","sourceLabel":"nvidia/Nemotron-3-Nano-Omni-30B-A3B-Reasoning-FP8 fp8 op de DGX Spark","sourceStatus":"complete","sourceTestId":"06-concurrency-stress","prefixCaching":false}],"openLoopResults":[{"id":"nemotron-3-nano-omni-30b-a3b-reasoning-fp8-office-random-4k","date":"2026-08-06","metrics":{"e2eMs":{"p50":39658.8,"p90":72540.53,"p95":78005.93,"p99":84900.08,"mean":40325.74},"tpotMs":{"p50":85.11,"p90":97.59,"p95":103.95,"p99":111.83,"mean":81.15},"ttftMs":{"p50":1279.84,"p90":2073.98,"p95":2276.15,"p99":2833.04,"mean":1360.91}},"sourceRun":"Test 07","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/nemotron-3/nemotron-3-nano-omni-30b-a3b-reasoning/fp8/","scenarioId":"office-random-4k","achievedRps":0.284,"sourceStatus":"complete","sourceTestId":"07-office-baseline","configuredRps":0.3,"prefixCaching":false,"totalRequests":200,"successfulRequests":200,"totalTokenThroughput":1296.56,"peakConcurrentRequests":24},{"id":"nemotron-3-nano-omni-30b-a3b-reasoning-fp8-sharegpt-replay","date":"2026-08-06","metrics":{"e2eMs":{"p50":4419.1,"p90":17847.03,"p95":22968.64,"p99":30055.96,"mean":7366.15},"tpotMs":{"p50":33.53,"p90":53.88,"p95":57.25,"p99":73.56,"mean":35.7},"ttftMs":{"p50":196.26,"p90":672.73,"p95":707.61,"p99":1118.86,"mean":277.93}},"sourceRun":"Test 08","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/nemotron-3/nemotron-3-nano-omni-30b-a3b-reasoning/fp8/","scenarioId":"sharegpt-replay","achievedRps":0.3,"sourceStatus":"complete","sourceTestId":"08-sharegpt","configuredRps":0.3,"prefixCaching":false,"totalRequests":250,"successfulRequests":250,"totalTokenThroughput":131.46,"peakConcurrentRequests":10},{"id":"nemotron-3-nano-omni-30b-a3b-reasoning-fp8-reasoning","date":"2026-08-06","metrics":{"e2eMs":{"p50":349402.46,"p90":500200.21,"p95":550173.34,"p99":574298.62,"mean":340560.74},"tpotMs":{"p50":83.89,"p90":95.01,"p95":98.18,"p99":101.66,"mean":83.34},"ttftMs":{"p50":815.82,"p90":1099.14,"p95":1423.1,"p99":1461.38,"mean":794.34}},"sourceRun":"Test 09","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/nemotron-3/nemotron-3-nano-omni-30b-a3b-reasoning/fp8/","scenarioId":"reasoning","achievedRps":0.072,"sourceStatus":"complete","sourceTestId":"09-reasoning","configuredRps":0.2,"prefixCaching":false,"totalRequests":50,"successfulRequests":50,"totalTokenThroughput":379.18,"peakConcurrentRequests":46},{"id":"nemotron-3-nano-omni-30b-a3b-reasoning-fp8-monday-peak-random-4k","date":"2026-08-06","metrics":{"e2eMs":{"p50":54257.77,"p90":115327.7,"p95":120227.83,"p99":126706.82,"mean":60249.72},"tpotMs":{"p50":129.04,"p90":147.25,"p95":284.85,"p99":1375.07,"mean":175.03},"ttftMs":{"p50":1225.34,"p90":1899.73,"p95":2964.89,"p99":3762.05,"mean":1333.19}},"sourceRun":"Test 10","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/nemotron-3/nemotron-3-nano-omni-30b-a3b-reasoning/fp8/","scenarioId":"monday-peak-random-4k","achievedRps":0.407,"sourceStatus":"complete","sourceTestId":"10-monday-peak","configuredRps":1.5,"prefixCaching":false,"totalRequests":300,"successfulRequests":198,"totalTokenThroughput":1827.85,"peakConcurrentRequests":28}]}