{"data":{"capabilities":{"chat":null,"reasoning":null,"tools":null,"vision":null},"description":null,"draft_launch":{"accelerator_backend":"nvidia","arguments":["--model","google/gemma-4-26B-A4B-it","--tensor-parallel-size","2","--host","0.0.0.0","--port","8000"],"container_port":8000,"entrypoint":null,"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":"0.26.0"},"facts":{"capabilities.chat":{"provenance":{"captured_at":"2026-08-27T06:04:13.773Z","sources":[{"captured_at":"2026-08-27T06:04:13.773Z","kind":"registry-enrichment","url":"https://github.com/0xSero/local-ai-registry"}]},"reason":"capability-not-verified","state":"unknown"},"capabilities.reasoning":{"provenance":{"captured_at":"2026-08-27T06:04:13.773Z","sources":[{"captured_at":"2026-08-27T06:04:13.773Z","kind":"registry-enrichment","url":"https://github.com/0xSero/local-ai-registry"}]},"reason":"capability-not-verified","state":"unknown"},"capabilities.tools":{"provenance":{"captured_at":"2026-08-27T06:04:13.773Z","sources":[{"captured_at":"2026-08-27T06:04:13.773Z","kind":"registry-enrichment","url":"https://github.com/0xSero/local-ai-registry"}]},"reason":"capability-not-verified","state":"unknown"},"capabilities.vision":{"provenance":{"captured_at":"2026-08-27T06:04:13.773Z","sources":[{"captured_at":"2026-08-27T06:04:13.773Z","kind":"registry-enrichment","url":"https://github.com/0xSero/local-ai-registry"}]},"reason":"capability-not-verified","state":"unknown"},"description":{"provenance":{"captured_at":"2026-08-27T06:04:13.773Z","sources":[{"captured_at":"2026-08-27T06:04:13.773Z","kind":"registry-enrichment","url":"https://github.com/0xSero/local-ai-registry"}]},"reason":"not-observed","state":"unknown"},"engine.graph_mode":{"provenance":{"captured_at":"2026-08-27T06:04:13.773Z","sources":[{"captured_at":"2026-08-27T06:04:13.773Z","kind":"registry-enrichment","url":"https://github.com/0xSero/local-ai-registry"}]},"reason":"runtime-detail-not-published","state":"unknown"},"metadata.localmaxxing.engine_flags.flashAttn":{"provenance":{"captured_at":"2026-08-27T06:04:13.773Z","sources":[{"captured_at":"2026-08-27T06:04:13.773Z","kind":"registry-enrichment","url":"https://github.com/0xSero/local-ai-registry"}]},"reason":"not-observed","state":"unknown"},"metadata.localmaxxing.engine_flags.gpuLayers":{"provenance":{"captured_at":"2026-08-27T06:04:13.773Z","sources":[{"captured_at":"2026-08-27T06:04:13.773Z","kind":"registry-enrichment","url":"https://github.com/0xSero/local-ai-registry"}]},"reason":"not-observed","state":"unknown"},"serving.kv_cache_tokens":{"provenance":{"captured_at":"2026-08-27T06:04:13.773Z","sources":[{"captured_at":"2026-08-27T06:04:13.773Z","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":"2026-08-27T06:04:13.773Z","sources":[{"captured_at":"2026-08-27T06:04:13.773Z","kind":"registry-enrichment","url":"https://github.com/0xSero/local-ai-registry"}]},"reason":"not-observed","state":"unknown"},"serving.max_context_tokens":{"provenance":{"captured_at":"2026-08-27T06:04:13.773Z","sources":[{"captured_at":"2026-08-27T06:04:13.773Z","kind":"registry-enrichment","url":"https://github.com/0xSero/local-ai-registry"}]},"reason":"not-observed","state":"unknown"}},"hardware_count":2,"hardware_id":"rtx-3090-24gb","id":"gemma-4-26b-a4b-int8-w8a16-rtx-3090-24gb-vllm-tp2","launch":{"container":{"captured_at":"2026-08-27T06:04:13.773Z","compose_file":null,"digest":null,"image":null,"reason":"reference-only-launch","runtime":null,"source":[{"captured_at":"2026-08-27T06:04:13.773Z","kind":"recipe-launch","url":"https://www.localmaxxing.com/en/runs/cms0ukqq600bys201w618abaa"}],"state":"none"},"kind":"reference","run_id":"cms0ukqq600bys201w618abaa","source":"localmaxxing","url":"https://www.localmaxxing.com/en/runs/cms0ukqq600bys201w618abaa"},"metadata":{"localmaxxing":{"base_lineage":"known","engine_flags":{"attentionBackend":"triton","flashAttn":null,"gpuLayers":null,"kvCacheDtype":"bf16","mtpEnabled":false,"specDecoding":true,"tensorParallel":2},"hardware_label":"2x RTX 3090 24GB","notes":"Second configuration of the same box: Eagle3 speculative decoding on top of the INT8 W8A16 run already submitted (145.2 t/s, no spec-decode). Drafter RedHatAI/gemma-4-26B-A4B-it-speculator.eagle3 (1.8 GB, speculators format), n=3. tokSOut 162.9 is prose, median of 3 (+12.2% over the non-speculative baseline); the same harness on a code prompt gives 206.2 t/s (+42.0%), since acceptance is far higher on syntactic text -- 46.2% accepted / 2.39 tokens per step on code vs 29.7% / 1.89 on prose. Method: stream:true, temperature 0, ignore_eos + min_tokens==max_tokens forcing exactly 400 tokens; rate = (n-1)/(t_last - t_first) so prefill and ramp are excluded, and n comes from usage.completion_tokens, NOT SSE chunk count -- under spec-decode vLLM packs each step's accepted tokens into one chunk, so counting chunks under-reports by the tokens-per-step factor. Aggregate decode 162.7 / 296.2 / 618.2 t/s at concurrency 1 / 2 / 4, vs 145.2 / 246.0 / 416.9 without spec-decode; --max-num-seqs 4 is a deliberate low-tenant setting, not a hardware limit. n=3 chosen over n=4 (faster single-stream on code at 218.8 t/s) because this endpoint serves parallel traffic and n=3 is clearly better under concurrency. Acceptance falls as num_speculative_tokens rises while tokens-per-step keeps climbing -- tokens/step is what predicts throughput, not accept rate. Costs vs non-speculative: KV pool 507,367 -> 374,279 tokens (still 1.43x full 262144 context), TTFT 38 -> 47 ms, prefill ~5% slower (9318 -> 8865 tok/s at 8k, 6434 at 32k). This same drafter was previously a large net loss on an NVFP4 build of this model; it wins here because int8_per_channel_weight_only quantizes weights only and leaves activations bf16, which is what the speculator was trained against. peakVramGb = 23240 MiB/card x2 at util 0.92. KV cache is bf16 because Gemma 4's interleaved attention routes to vLLM's Triton backend, which needs SM89+ for native FP8 -- unavailable on Ampere. 2x RTX 3090 NVLinked (NV4).","revision_unpinned":true,"run_id":"cms0ukqq600bys201w618abaa"}},"model_instance_id":"google-gemma-4-26b-a4b-it--int8-w8a16","provenance":{"captured_at":"2026-08-27T06:04:13.773Z","sources":[{"captured_at":"2026-08-27T06:04:13.773Z","kind":"normalized-recipe","url":"https://www.localmaxxing.com/en/runs/cms0ukqq600bys201w618abaa"}]},"recipe_source":"localmaxxing","schema_version":"local-ai-registry/v1","serving":{"kv_cache_tokens":null,"max_concurrency":null,"max_context_tokens":null,"tensor_parallel":2},"speed_sweep_ids":["gemma-4-26b-a4b-int8-w8a16-rtx-3090-24gb-vllm-tp2-sweep"],"status":"candidate","huggingface":{"link_type":"repository","provenance":{"captured_at":"2026-08-27T06:04:13.773Z","sources":[{"captured_at":"2026-08-27T06:04:13.773Z","kind":"huggingface-api","url":"https://huggingface.co/api/models/google/gemma-4-26B-A4B-it"}]},"reason":"hf-api-confirmed-public","repository":"google/gemma-4-26B-A4B-it","status":"known","url":"https://huggingface.co/google/gemma-4-26B-A4B-it"},"registry":{"launchable":false,"speed_evidence":{"available":true,"count":1,"speed_sweep_ids":["gemma-4-26b-a4b-int8-w8a16-rtx-3090-24gb-vllm-tp2-sweep"],"detail_urls":["/api/v1/speed-sweep/gemma-4-26b-a4b-int8-w8a16-rtx-3090-24gb-vllm-tp2-sweep"]},"runtime":"reference"},"relationships":{"hardware":{"api":"/api/v1/hardware/rtx-3090-24gb","href":"/hardware/rtx-3090-24gb","id":"rtx-3090-24gb","name":"GeForce RTX 3090"},"model":{"api":"/api/v1/models/gemma-4-26b-a4b","href":"/models/gemma-4-26b-a4b","id":"gemma-4-26b-a4b","name":"gemma-4-26B-A4B"},"model_instance":{"api":"/api/v1/model-instances/google-gemma-4-26b-a4b-it--int8-w8a16","href":"/model-instances/google-gemma-4-26b-a4b-it--int8-w8a16","id":"google-gemma-4-26b-a4b-it--int8-w8a16","name":"google/gemma-4-26B-A4B-it"},"speed_sweep":[{"api":"/api/v1/speed-sweep/gemma-4-26b-a4b-int8-w8a16-rtx-3090-24gb-vllm-tp2-sweep","href":"/speed-sweep/gemma-4-26b-a4b-int8-w8a16-rtx-3090-24gb-vllm-tp2-sweep","id":"gemma-4-26b-a4b-int8-w8a16-rtx-3090-24gb-vllm-tp2-sweep"}]}},"meta":{"source":"registry"}}