{"data":{"capabilities":{"chat":true,"reasoning":false,"tools":true,"vision":false},"description":"Validated one-B70 Qwen3.8-27B Q4_K_M llama.cpp SYCL runtime using the published digest-pinned image, one 131,072-token slot, correct completion and tool-call acceptance, and six accepted or historical speed points through 128K context.","engine":{"graph_mode":"disabled at compile time (-DGGML_SYCL_GRAPH=OFF)","name":"llama-cpp","version":"4302fb59969a5d8cf9f8e5f55fdd4506d0ed2126+b70-patches"},"facts":{"engine.graph_mode":{"provenance":{"captured_at":"2026-09-01T09:21:24Z","sources":[{"captured_at":"2026-09-01T09:21:24Z","kind":"recipe-source","url":"https://github.com/0xSero/qwen38-b70/blob/e8738535c8914a4fe6a54235754e27eeb8078601/Dockerfile"}]},"reason":"published-dockerfile-builds-llama-cpp-with-ggml-sycl-graph-off","state":"known"},"launch.container.digest":{"provenance":{"captured_at":"2026-09-01T09:21:24Z","sources":[{"captured_at":"2026-09-01T09:21:24Z","kind":"oci-registry-manifest","url":"https://github.com/0xSero/qwen38-b70/pkgs/container/qwen38-b70"}]},"reason":"docker-manifest-inspect-verified-the-published-oci-index-and-linux-amd64-manifest","state":"known"},"serving.kv_cache_tokens":{"provenance":{"captured_at":"2026-09-01T09:21:24Z","sources":[{"captured_at":"2026-09-01T09:21:24Z","kind":"recipe-source","url":"https://github.com/0xSero/qwen38-b70/blob/e8738535c8914a4fe6a54235754e27eeb8078601/README.md"}]},"reason":"published-single-gpu-launch-defines-one-131072-token-slot","state":"known"},"serving.max_context_tokens":{"provenance":{"captured_at":"2026-09-01T09:21:24Z","sources":[{"captured_at":"2026-09-01T09:21:24Z","kind":"recipe-source","url":"https://github.com/0xSero/qwen38-b70/blob/e8738535c8914a4fe6a54235754e27eeb8078601/README.md"}]},"reason":"published-single-gpu-launch-and-performance-table-cover-the-131072-token-envelope","state":"known"}},"hardware_count":1,"hardware_id":"intel-arc-pro-b70-32gb","id":"qwen38-q4km-arcb70-llamacpp-tp1","launch":{"accelerator_backend":"intel-xpu","arguments":["--device","SYCL0","--gpu-layers","99","--flash-attn","on","--batch-size","8192","--ubatch-size","8192","--cache-type-k","f16","--cache-type-v","f16","--ctx-size","131072","--parallel","1"],"container":{"captured_at":"2026-08-27T06:04:13.773Z","compose_file":null,"digest":"sha256:37219eaeeb2e8398f59e427ec0c723bbe2eeb8ec59e6b154ea63c4172238c342","image":"ghcr.io/0xsero/qwen38-b70@sha256:37219eaeeb2e8398f59e427ec0c723bbe2eeb8ec59e6b154ea63c4172238c342","reason":"image-reference-in-launch","runtime":"docker","source":[{"captured_at":"2026-08-27T06:04:13.773Z","kind":"recipe-launch","url":"https://github.com/0xSero/local-ai-registry"}],"state":"digest-pinned"},"container_port":8010,"environment":{"BATCH":"8192","CTX_SIZE_OVERRIDE":"131072","ENABLE_MTP":"0","ENABLE_VISION":"0","GPU_COUNT":"1","MODELS_DIR":"/models","PARALLEL":"1","THREADS":"16","UBATCH":"8192"},"host_port":8010,"image":"ghcr.io/0xsero/qwen38-b70-attested@sha256:6ce7771eaa155e31916ac3144bd3aa34bed6584b9a43052a8f6b583b9c8f5db1","kind":"docker","mounts":[{"read_only":true,"source":"/dev/dri/by-path","target":"/dev/dri/by-path"},{"read_only":true,"source":"${MODEL_ROOT}/qwen38-b70","target":"/models"}],"network_mode":"bridge","provenance":{"attestation":"gh attestation verify oci://ghcr.io/0xsero/qwen38-b70-attested@sha256:6ce7771eaa155e31916ac3144bd3aa34bed6584b9a43052a8f6b583b9c8f5db1 -o 0xSero","dockerfile":"https://github.com/0xSero/qwen38-b70/blob/main/Dockerfile","kind":"self-built-attested","source":"https://github.com/0xSero/qwen38-b70","workflow":"https://github.com/0xSero/qwen38-b70/actions/runs/33567506333"},"shm_size":"8g"},"metadata":{"runtime_identity":{"model_revision":"0669b98607d47046c7c2b3f801011d54a08cfccf","published_oci_index_digest":"sha256:37219eaeeb2e8398f59e427ec0c723bbe2eeb8ec59e6b154ea63c4172238c342","published_oci_manifest_digest":"sha256:f537fc6d1cf93d7d80253e23e11cbd6a8e821382c87ac050f025743643f49629","slot_count":1,"source_commit":"e8738535c8914a4fe6a54235754e27eeb8078601"}},"model_instance_id":"ggml-org-qwen3-8-27b-gguf--q4-k-m","provenance":{"captured_at":"2026-08-27T06:04:13.773Z","sources":[{"captured_at":"2026-08-27T06:04:13.773Z","kind":"normalized-recipe","url":"https://github.com/0xSero/local-ai-registry"}]},"recipe_source":"0xsero","recommended":true,"schema_version":"local-ai-registry/v1","serving":{"kv_cache_tokens":131072,"max_concurrency":1,"max_context_tokens":131072,"tensor_parallel":1},"speed_sweep_ids":["qwen38-q4km-arcb70-llamacpp-tp1-sweep"],"status":"validated","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/ggml-org/Qwen3.8-27B-GGUF"}]},"reason":"hf-api-confirmed-public","repository":"ggml-org/Qwen3.8-27B-GGUF","status":"known","url":"https://huggingface.co/ggml-org/Qwen3.8-27B-GGUF"},"registry":{"launchable":true,"speed_evidence":{"available":true,"count":1,"speed_sweep_ids":["qwen38-q4km-arcb70-llamacpp-tp1-sweep"],"detail_urls":["/api/v1/speed-sweep/qwen38-q4km-arcb70-llamacpp-tp1-sweep"]},"runtime":"docker"},"relationships":{"hardware":{"api":"/api/v1/hardware/intel-arc-pro-b70-32gb","href":"/hardware/intel-arc-pro-b70-32gb","id":"intel-arc-pro-b70-32gb","name":"Intel Arc Pro B70"},"model":{"api":"/api/v1/models/qwen3-8-27b","href":"/models/qwen3-8-27b","id":"qwen3-8-27b","name":"Qwen3.8-27B"},"model_instance":{"api":"/api/v1/model-instances/ggml-org-qwen3-8-27b-gguf--q4-k-m","href":"/model-instances/ggml-org-qwen3-8-27b-gguf--q4-k-m","id":"ggml-org-qwen3-8-27b-gguf--q4-k-m","name":"ggml-org/Qwen3.8-27B-GGUF"},"speed_sweep":[{"api":"/api/v1/speed-sweep/qwen38-q4km-arcb70-llamacpp-tp1-sweep","href":"/speed-sweep/qwen38-q4km-arcb70-llamacpp-tp1-sweep","id":"qwen38-q4km-arcb70-llamacpp-tp1-sweep"}]}},"meta":{"source":"registry"}}