{"name":"Qwen-3.6-35B-A3B","arena":{"id":"qwen-3-6-35b-a3b-nvfp4","name":"Qwen-3.6 35B-A3B","vram":24,"params":35,"vendor":"alibaba","quality":{"coding":{"bench":"LiveCodeBench v6","value":80.4},"science":{"bench":"GPQA-Diamond","value":86},"knowledge":{"bench":"MMLU-Pro","value":85.2}},"contextK":256,"baseModelId":"qwen-3-6-35b-a3b"},"notes":"Model-ID: RedHatAI/Qwen3.6-35B-A3B-NVFP4. Container: vllm/vllm-openai:v0.26.0-aarch64-cu129-ubuntu2404. llama-benchy 0.4.0. Bronrun: 2026-08-14T06:58:43+02:00. KV-cache fp8; max_model_len 131072; gpu_memory_utilization 0,9. Prefix caching uit; async scheduling aan. Extra serverflags: --language-model-only --reasoning-parser qwen3 --trust-remote-code --moe-backend marlin --linear-backend marlin. Profielvariabelen: VLLM_USE_FLASHINFER_MOE_FP4=0. 11 van de 11 tests hebben een meting die volgens de huidige Arena-regels meetelt. De sanity-check is geslaagd. Dit is een controle op een kort gegenereerd antwoord, geen inhoudelijke evaluatie van de benchmarktaken.","order":360,"engine":"vLLM v0.26.0","verdict":"Gemeten op 2026-08-14 met vLLM v0.26.0. In de chattest met tien gelijktijdige verzoeken haalt dit profiel 20,27 tokens/s per gebruiker, met gemiddeld 1,21 seconden tot het eerste token. Bij 25k context stijgt die gemiddelde wachttijd naar 26,28 seconden. Deze runs meten snelheid en wachttijd, niet de kwaliteit van de antwoorden.","hardware":"DGX Spark, NVIDIA GB10, 128 GB unified memory","modelUrl":"https://huggingface.co/Qwen/Qwen3.6-35B-A3B","provider":"Alibaba","precision":"NVFP4","parameters":"35B (3B active)","provenance":{"modelId":"RedHatAI/Qwen3.6-35B-A3B-NVFP4","runPath":"results/qwen-3.6/qwen-3.6-35b-a3b/nvfp4","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":"passed"},"sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/qwen-3.6/qwen-3.6-35b-a3b/nvfp4/","servedName":"qwen3.6-35b-a3b-nvfp4","generatedAt":"2026-08-14T06:58:43+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":"qwen-3-6-35b-a3b-nvfp4-chat-1k","date":"2026-08-14","runs":3,"metrics":{"ttfrSec":{"mean":1.214,"stddev":0.455},"decodeTokensPerSecondTotal":{"mean":196.13,"stddev":2.16},"prefillTokensPerSecondTotal":{"mean":5281.78,"stddev":169.38},"decodeTokensPerSecondPerUser":{"mean":20.27,"stddev":0.35}},"sourceRun":"Test 01","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/qwen-3.6/qwen-3.6-35b-a3b/nvfp4/","workloadId":"chat-1k","sourceLabel":"RedHatAI/Qwen3.6-35B-A3B-NVFP4 nvfp4 op de DGX Spark","sourceStatus":"complete","sourceTestId":"01-chat","prefixCaching":false},{"id":"qwen-3-6-35b-a3b-nvfp4-rag-8k","date":"2026-08-14","runs":3,"metrics":{"ttfrSec":{"mean":7.729,"stddev":3.875},"decodeTokensPerSecondTotal":{"mean":135.11,"stddev":0.79},"prefillTokensPerSecondTotal":{"mean":5433.02,"stddev":46.21},"decodeTokensPerSecondPerUser":{"mean":16.59,"stddev":1.9}},"sourceRun":"Test 02","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/qwen-3.6/qwen-3.6-35b-a3b/nvfp4/","workloadId":"rag-8k","sourceLabel":"RedHatAI/Qwen3.6-35B-A3B-NVFP4 nvfp4 op de DGX Spark","sourceStatus":"complete","sourceTestId":"02-rag-8k","prefixCaching":false},{"id":"qwen-3-6-35b-a3b-nvfp4-long-output-agents","date":"2026-08-14","runs":3,"metrics":{"ttfrSec":{"mean":0.466,"stddev":0.078},"decodeTokensPerSecondTotal":{"mean":176.06,"stddev":3.32},"prefillTokensPerSecondTotal":{"mean":4275.33,"stddev":36.66},"decodeTokensPerSecondPerUser":{"mean":22.4,"stddev":1.54}},"sourceRun":"Test 03","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/qwen-3.6/qwen-3.6-35b-a3b/nvfp4/","workloadId":"long-output-agents","sourceLabel":"RedHatAI/Qwen3.6-35B-A3B-NVFP4 nvfp4 op de DGX Spark","sourceStatus":"complete","sourceTestId":"03-long-output","prefixCaching":false},{"id":"qwen-3-6-35b-a3b-nvfp4-multi-turn-office","date":"2026-08-14","runs":3,"metrics":{"ttfrSec":{"mean":2.208,"stddev":0.998},"decodeTokensPerSecondTotal":{"mean":183.61,"stddev":0.9},"prefillTokensPerSecondTotal":{"mean":5351.18,"stddev":351.23},"decodeTokensPerSecondPerUser":{"mean":19.58,"stddev":0.66}},"sourceRun":"Test 04","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/qwen-3.6/qwen-3.6-35b-a3b/nvfp4/","workloadId":"multi-turn-office","sourceLabel":"RedHatAI/Qwen3.6-35B-A3B-NVFP4 nvfp4 op de DGX Spark","sourceStatus":"complete","sourceTestId":"04-multi-turn","prefixCaching":false},{"id":"qwen-3-6-35b-a3b-nvfp4-large-context-25k","date":"2026-08-14","runs":3,"metrics":{"ttfrSec":{"mean":26.277,"stddev":13.603},"decodeTokensPerSecondTotal":{"mean":45.98,"stddev":0.35},"prefillTokensPerSecondTotal":{"mean":4768.86,"stddev":36.03},"decodeTokensPerSecondPerUser":{"mean":9.41,"stddev":4.5}},"sourceRun":"Test 05","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/qwen-3.6/qwen-3.6-35b-a3b/nvfp4/","workloadId":"large-context-25k","sourceLabel":"RedHatAI/Qwen3.6-35B-A3B-NVFP4 nvfp4 op de DGX Spark","sourceStatus":"complete","sourceTestId":"05-big-context","prefixCaching":false},{"id":"qwen-3-6-35b-a3b-nvfp4-concurrency-stress-25k","date":"2026-08-14","runs":3,"metrics":{"ttfrSec":{"mean":50.719,"stddev":28.111},"decodeTokensPerSecondTotal":{"mean":47.57,"stddev":0.05},"prefillTokensPerSecondTotal":{"mean":4657.34,"stddev":2},"decodeTokensPerSecondPerUser":{"mean":5.91,"stddev":3.76}},"sourceRun":"Test 06","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/qwen-3.6/qwen-3.6-35b-a3b/nvfp4/","workloadId":"concurrency-stress-25k","sourceLabel":"RedHatAI/Qwen3.6-35B-A3B-NVFP4 nvfp4 op de DGX Spark","sourceStatus":"complete","sourceTestId":"06-concurrency-stress","prefixCaching":false}],"openLoopResults":[{"id":"qwen-3-6-35b-a3b-nvfp4-office-random-4k","date":"2026-08-14","metrics":{"e2eMs":{"p50":26092.46,"p90":48332.46,"p95":54772.59,"p99":60292.97,"mean":27369.12},"tpotMs":{"p50":54.38,"p90":67.81,"p95":73.1,"p99":82.76,"mean":54.45},"ttftMs":{"p50":1083.14,"p90":1970.39,"p95":2388.6,"p99":3232.88,"mean":1183.08}},"sourceRun":"Test 07","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/qwen-3.6/qwen-3.6-35b-a3b/nvfp4/","scenarioId":"office-random-4k","achievedRps":0.289,"sourceStatus":"complete","sourceTestId":"07-office-baseline","configuredRps":0.3,"prefixCaching":false,"totalRequests":200,"successfulRequests":200,"totalTokenThroughput":1316.83,"peakConcurrentRequests":16},{"id":"qwen-3-6-35b-a3b-nvfp4-sharegpt-replay","date":"2026-08-14","metrics":{"e2eMs":{"p50":3957.23,"p90":15309.85,"p95":18025.24,"p99":26216.6,"mean":6332.03},"tpotMs":{"p50":28.76,"p90":38.05,"p95":39.23,"p99":42.78,"mean":29.57},"ttftMs":{"p50":144.34,"p90":226.18,"p95":243.98,"p99":292.18,"mean":154.15}},"sourceRun":"Test 08","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/qwen-3.6/qwen-3.6-35b-a3b/nvfp4/","scenarioId":"sharegpt-replay","achievedRps":0.299,"sourceStatus":"complete","sourceTestId":"08-sharegpt","configuredRps":0.3,"prefixCaching":false,"totalRequests":250,"successfulRequests":250,"totalTokenThroughput":131.56,"peakConcurrentRequests":9},{"id":"qwen-3-6-35b-a3b-nvfp4-reasoning","date":"2026-08-14","metrics":{"e2eMs":{"p50":339824.01,"p90":508298.62,"p95":553089.67,"p99":572418.07,"mean":334818.56},"tpotMs":{"p50":82.27,"p90":92.1,"p95":93.41,"p99":97.65,"mean":80.6},"ttftMs":{"p50":379.54,"p90":547.89,"p95":588.62,"p99":616.45,"mean":388.68}},"sourceRun":"Test 09","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/qwen-3.6/qwen-3.6-35b-a3b/nvfp4/","scenarioId":"reasoning","achievedRps":0.07,"sourceStatus":"complete","sourceTestId":"09-reasoning","configuredRps":0.2,"prefixCaching":false,"totalRequests":50,"successfulRequests":50,"totalTokenThroughput":372.8,"peakConcurrentRequests":44},{"id":"qwen-3-6-35b-a3b-nvfp4-monday-peak-random-4k","date":"2026-08-14","metrics":{"e2eMs":{"p50":52260.67,"p90":95488.85,"p95":99443.9,"p99":104839.21,"mean":53557.64},"tpotMs":{"p50":106.41,"p90":115.14,"p95":119.34,"p99":135.49,"mean":106.03},"ttftMs":{"p50":1074.28,"p90":1897.18,"p95":2520.76,"p99":6501.61,"mean":1268.91}},"sourceRun":"Test 10","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/qwen-3.6/qwen-3.6-35b-a3b/nvfp4/","scenarioId":"monday-peak-random-4k","achievedRps":0.449,"sourceStatus":"complete","sourceTestId":"10-monday-peak","configuredRps":1.5,"prefixCaching":false,"totalRequests":300,"successfulRequests":300,"totalTokenThroughput":2023.4,"peakConcurrentRequests":28}],"rateSweep":{"date":"2026-08-14","capacity":{"slo2000":{"outputTps":48.1,"ttftP95Ms":1986,"achievedRps":0.098,"configuredRps":0.1,"peakConcurrent":6},"slo5000":{"outputTps":272,"ttftP95Ms":3821,"achievedRps":0.557,"configuredRps":0.7,"peakConcurrent":73},"slo10000":{"outputTps":272,"ttftP95Ms":3821,"achievedRps":0.557,"configuredRps":0.7,"peakConcurrent":73}},"sourceRun":"Test 11","sourceUrl":"https://github.com/djangodevreng/dgx-spark-benchmarks/tree/a948cec23b054c62fef595bad3cdb284066922ac/results/qwen-3.6/qwen-3.6-35b-a3b/nvfp4/","abortedAtRps":null,"queueKneeRps":0.7,"sourceStatus":"complete","sourceTestId":"11-rate-sweep","totalRequests":850,"successfulRequests":850}}