{"data":{"capabilities":{"chat":null,"reasoning":null,"tools":null,"vision":null},"description":"Observed LocalMaxxing leaderboard run. Evidence for compatibility, not an executable launch contract.","draft_launch":{"accelerator_backend":"nvidia","arguments":["serve","--model","sakamakismile/Qwen3.6-27B-NVFP4","--tensor-parallel-size","1","--host","0.0.0.0","--port","8000","--max-model-len","262144"],"container_port":8000,"entrypoint":"vllm","environment":{},"host_port":8000,"image":"vllm/vllm-openai@sha256:0a51ea5b4ae2dc5d81890e5173f54203d2a3ae0cfffe51b8fd2afd4391bfd967","ipc":"host","kind":"docker","mounts":[{"read_only":false,"source":"~/.cache/huggingface","target":"/root/.cache/huggingface"}],"shm_size":"16g","synthesized":{"generated_at":"2026-08-31T22:12:17Z","image_provenance":"deepseek-fp8-rtx-pro-6000-blackwell-96gb-vllm-tp1","template":"vllm-openai-v1"}},"engine":{"graph_mode":null,"name":"vllm","version":"0.19.2rc1.dev213+g9558f4390"},"facts":{"capabilities.chat":{"provenance":{"captured_at":"2026-08-30T09:26:09Z","sources":[{"captured_at":"2026-08-30T09:26:09Z","kind":"registry-enrichment","url":"https://github.com/0xSero/local-ai-registry"}]},"reason":"capability-not-verified","state":"unknown"},"capabilities.reasoning":{"provenance":{"captured_at":"2026-08-30T09:26:09Z","sources":[{"captured_at":"2026-08-30T09:26:09Z","kind":"registry-enrichment","url":"https://github.com/0xSero/local-ai-registry"}]},"reason":"capability-not-verified","state":"unknown"},"capabilities.tools":{"provenance":{"captured_at":"2026-08-30T09:26:09Z","sources":[{"captured_at":"2026-08-30T09:26:09Z","kind":"registry-enrichment","url":"https://github.com/0xSero/local-ai-registry"}]},"reason":"capability-not-verified","state":"unknown"},"capabilities.vision":{"provenance":{"captured_at":"2026-08-30T09:26:09Z","sources":[{"captured_at":"2026-08-30T09:26:09Z","kind":"registry-enrichment","url":"https://github.com/0xSero/local-ai-registry"}]},"reason":"capability-not-verified","state":"unknown"},"draft_launch.entrypoint":{"provenance":{"captured_at":"2026-09-01T01:41:26Z","sources":[{"captured_at":"2026-09-01T01:41:26Z","kind":"container-config","url":"https://registry-1.docker.io/v2/vllm/vllm-openai/manifests/sha256:0a51ea5b4ae2dc5d81890e5173f54203d2a3ae0cfffe51b8fd2afd4391bfd967"}]},"reason":"linux-amd64-container-config-entrypoint","state":"known"},"engine.graph_mode":{"provenance":{"captured_at":"2026-08-30T09:26:09Z","sources":[{"captured_at":"2026-08-30T09:26:09Z","kind":"registry-enrichment","url":"https://github.com/0xSero/local-ai-registry"}]},"reason":"runtime-detail-not-published","state":"unknown"},"serving.kv_cache_tokens":{"provenance":{"captured_at":"2026-08-30T09:26:09Z","sources":[{"captured_at":"2026-08-30T09:26:09Z","kind":"registry-enrichment","url":"https://github.com/0xSero/local-ai-registry"}]},"reason":"kv-cache-capacity-not-published","state":"unknown"},"serving.max_concurrency":{"provenance":{"captured_at":"2026-08-31T23:03:15Z","sources":[{"captured_at":"2026-08-31T23:03:15Z","kind":"registry-derived","url":"https://www.localmaxxing.com/en/runs/cmogn5c7w0003l704v5lfqrqu"}]},"reason":"server-capacity-derived-from-source-evidence","state":"known"}},"hardware_count":1,"hardware_id":"dgx-spark-gb10-128gb","id":"qwen3-6-27b-nvfp4-dgx-spark-gb10-128gb-vllm-tp1-v5lfqrqu","launch":{"container":{"captured_at":"2026-08-30T09:26:09Z","compose_file":null,"digest":null,"image":null,"reason":"reference-only-launch","runtime":null,"source":[{"captured_at":"2026-08-30T09:26:09Z","kind":"recipe-launch","url":"https://www.localmaxxing.com/en/runs/cmogn5c7w0003l704v5lfqrqu"}],"state":"none"},"kind":"reference","run_id":"cmogn5c7w0003l704v5lfqrqu","source":"localmaxxing","url":"https://www.localmaxxing.com/en/runs/cmogn5c7w0003l704v5lfqrqu"},"metadata":{"localmaxxing":{"backend":"cuda","batch_size":1,"hardware_label":"GB10 Grace Blackwell","notes":"Final corrected submission (replaces cmoglz0ux0004lb04zo1dhhhb and cmogm80ri0001lb04hwhu9uzx). Target weights: sakamakismile/Qwen3.6-27B-NVFP4 (community NVFP4 quant of Qwen/Qwen3.6-27B). Drafter: z-lab/Qwen3.6-27B-DFlash (DFlash spec decoding, num_speculative_tokens=15, thinking-ON). Hardware: DGX Spark GB10 (128 GB unified, peak 117 GiB measured). n=30 trials @ pp=128, tg=128, depth=0, concurrency=1, warm-pass (cold-start sample dropped): tokSOut median=32.83 (mean=40.10, std=15.63), ttfr median=268ms (mean=274). Median is the honest headline; DFlash decode std is ~40% of mean due to acceptance fluctuating with prompt content. Mean acceptance length tau=4.11 (per-position [0.75, 0.52, 0.38, ...]). Speedup vs Qwen3.6-27B-FP8 autoregressive baseline on same hardware: 4.18x (32.83 / 7.85). First DGX Spark + spec-decoding entry, first NVFP4 entry leaderboard-wide. REQUIRES the DFlash layer-tap off-by-one vLLM source patch (without it acceptance collapses ~80% -> ~3%). llama-benchy 0.3.6 (PR #11 SSE-counting fix is upstream as of 0.3.6). Full Docker reproduction: https://github.com/my-other-github-account/qwen36-dflash-spark-repro By @banana_baeee.","observed_command":"# Reproduction — see https://github.com/my-other-github-account/qwen36-dflash-spark-repro\n# Download models on host:\nhf download sakamakismile/Qwen3.6-27B-NVFP4 --local-dir ~/models/Qwen3.6-27B-NVFP4\nhf download z-lab/Qwen3.6-27B-DFlash       --local-dir ~/models/Qwen3.6-27B-DFlash\n\n# Build & run:\ndocker build -t qwen36-dflash-spark .\ndocker run -d --name qwen36-dflash --runtime=nvidia --gpus all --network=host \\\n  -v ~/models:/models:ro \\\n  -e THINK_KWARGS='{\"enable_thinking\": true}' \\\n  qwen36-dflash-spark\n\n# Bench (pp=128 / tg=128 / depth=0 / c=1 / n=30) — produces the headline number:\ndocker run --rm --network=host -v ~/models:/models:ro -v $(pwd):/out \\\n  --entrypoint bash qwen36-dflash-spark \\\n  -c 'OUT=/out/result.json bash /repro/scripts/bench.sh'\n\n# Underlying vllm serve invocation (from launch_server.sh):\nvllm serve /models/Qwen3.6-27B-NVFP4 \\\n  --served-model-name qwen36-27b \\\n  --host 0.0.0.0 --port 8000 \\\n  --tensor-parallel-size 1 \\\n  --gpu-memory-utilization 0.92 \\\n  --max-model-len 262144 \\\n  --max-num-batched-tokens 4096 \\\n  --max-num-seqs 1 \\\n  --trust-remote-code \\\n  --load-format fastsafetensors \\\n  --attention-backend flash_attn \\\n  --enable-prefix-caching \\\n  --speculative-config '{\"method\":\"dflash\",\"num_speculative_tokens\":15,\"model\":\"/models/Qwen3.6-27B-DFlash\"}' \\\n  --default-chat-template-kwargs '{\"enable_thinking\": true}' \\\n  --seed 0\n\n# Required vLLM source patch: patches/apply_dflash_off_by_one.sh (DFlash layer-tap +1).","run_id":"cmogn5c7w0003l704v5lfqrqu","tokenized":{"fidelity":"lossy"}}},"model_instance_id":"sakamakismile-qwen3-6-27b-nvfp4--nvfp4","provenance":{"captured_at":"2026-08-30T09:26:09Z","sources":[{"captured_at":"2026-08-30T09:26:09Z","kind":"normalized-recipe","url":"https://www.localmaxxing.com/en/runs/cmogn5c7w0003l704v5lfqrqu"}]},"recipe_source":"localmaxxing","schema_version":"local-ai-registry/v1","serving":{"kv_cache_tokens":null,"max_concurrency":1,"max_context_tokens":262144,"tensor_parallel":1},"speed_sweep_ids":["qwen3-6-27b-nvfp4-dgx-spark-gb10-128gb-vllm-tp1-v5lfqrqu-sweep"],"status":"candidate","huggingface":{"link_type":"repository","provenance":{"captured_at":"2026-08-30T09:10:02Z","sources":[{"captured_at":"2026-08-30T09:10:02Z","kind":"huggingface-api","url":"https://huggingface.co/api/models/sakamakismile/Qwen3.6-27B-NVFP4"}]},"reason":"hf-api-confirmed-public","repository":"sakamakismile/Qwen3.6-27B-NVFP4","status":"known","url":"https://huggingface.co/sakamakismile/Qwen3.6-27B-NVFP4"},"registry":{"launchable":false,"speed_evidence":{"available":true,"count":1,"speed_sweep_ids":["qwen3-6-27b-nvfp4-dgx-spark-gb10-128gb-vllm-tp1-v5lfqrqu-sweep"],"detail_urls":["/api/v1/speed-sweep/qwen3-6-27b-nvfp4-dgx-spark-gb10-128gb-vllm-tp1-v5lfqrqu-sweep"]},"runtime":"reference"},"relationships":{"hardware":{"api":"/api/v1/hardware/dgx-spark-gb10-128gb","href":"/hardware/dgx-spark-gb10-128gb","id":"dgx-spark-gb10-128gb","name":"NVIDIA DGX Spark GB10"},"model":{"api":"/api/v1/models/qwen3-6-27b","href":"/models/qwen3-6-27b","id":"qwen3-6-27b","name":"Qwen3.6-27B"},"model_instance":{"api":"/api/v1/model-instances/sakamakismile-qwen3-6-27b-nvfp4--nvfp4","href":"/model-instances/sakamakismile-qwen3-6-27b-nvfp4--nvfp4","id":"sakamakismile-qwen3-6-27b-nvfp4--nvfp4","name":"sakamakismile/Qwen3.6-27B-NVFP4"},"speed_sweep":[{"api":"/api/v1/speed-sweep/qwen3-6-27b-nvfp4-dgx-spark-gb10-128gb-vllm-tp1-v5lfqrqu-sweep","href":"/speed-sweep/qwen3-6-27b-nvfp4-dgx-spark-gb10-128gb-vllm-tp1-v5lfqrqu-sweep","id":"qwen3-6-27b-nvfp4-dgx-spark-gb10-128gb-vllm-tp1-v5lfqrqu-sweep"}]}},"meta":{"source":"registry"}}