{"command":"eval benchmark context artifact --run-id swift-tp2-r3-context","context":{"command":null,"command_host":null,"command_runtime":null,"controller_endpoint":null,"controller_endpoint_kind":null,"execution_host":null,"execution_runtime":null,"gpu_role":null,"gpu_uuid":null,"overlay":null,"resource":null,"resource_endpoint":null,"resource_endpoint_kind":null,"resource_host":null,"resource_runtime":null,"target":null,"topology":null,"transport":null},"data":"{\"data\": {\"created_at\": \"2026-09-26T18:04:59.979699Z\", \"failure\": null, \"partial\": false, \"provenance\": {\"endpoint\": {\"base_url\": \"http://127.0.0.1:39146/v1\", \"model\": \"swift15-flash-next-nvfp4-tp2-327k-r3\"}, \"finished_at\": \"2026-09-26T18:04:59.979628Z\", \"started_at\": \"2026-09-26T18:01:34.268586Z\", \"submitted_at\": \"2026-09-26T17:50:34.239063+00:00\", \"worker\": {\"id\": \"local-benchmark\"}}, \"results\": {\"evidence\": {\"completeness\": \"completed\", \"created_at\": \"2026-09-26T18:04:59.970473Z\", \"evidence_kind\": \"measured\", \"failure\": null, \"identities\": {\"concurrency\": {\"requests\": 1, \"workers\": 1}, \"context\": {\"configured_tokens\": 327680}, \"dataset\": {\"name\": \"deterministic-native\", \"profile_sha256\": \"fce0e6e1e7497585bdfc6f76507d3baaff99a6b4f174d841d268de20c68de7f6\"}, \"hardware\": {\"gpu\": \"2x RTX PRO 6000 Blackwell Max-Q\", \"tensor_parallel_size\": 2, \"vram_mib_per_gpu\": 97887}, \"harnesses\": {\"native\": {\"schema\": \"anvil-serving.context/v1\"}}, \"image\": {\"status\": \"not-applicable\"}, \"model\": {\"repository\": \"ukisai/Swift-1.5-Qwen3.8-Flash-Next-NVFP4\", \"revision\": \"3ff0520224f264a2d0ac4ab56ece8f2f13aadb38\", \"served_name\": \"swift15-flash-next-nvfp4-tp2-327k-r3\"}, \"runtime\": {\"engine_revision\": \"ced6857afa0ea7b2e3f0846a62e1394e90f15607\", \"image_id\": \"sha256:8a69ffad015f138d7170c4ddc429e230a3bc1c1719f67e14324749df200a4b90\"}, \"served_model\": {\"configured_context\": 327680, \"id\": \"swift15-flash-next-nvfp4-tp2-327k-r3\"}, \"topology\": {\"kind\": \"co-resident-benchmark-client\", \"worker_id\": \"local-benchmark\"}}, \"promotion\": {\"authorized\": false, \"message\": \"Benchmark evidence does not authorize model promotion; promotion is a separate human decision.\"}, \"run\": {\"ownership_id\": \"swift-tp2-20260926\", \"profile\": \"scout\", \"run_id\": \"swift-tp2-r3-context\", \"spec_sha256\": \"8fba4c78bd0a714f3df9c0a35caf09c4daa9ef015bebef3442b6bd16d085ab57\", \"suite\": \"context\"}, \"schema\": \"anvil-serving.benchmark-evidence/v1\", \"stages\": [{\"evidence\": [{\"bytes\": 542, \"path\": \"evidence/0-assets.json\", \"sha256\": \"08cfb26fd5bd2d13b2cf09b5d7bb7c520f7b54357fed195f716721256db4f769\"}], \"name\": \"asset_preparation\", \"sequence\": 0, \"status\": \"completed\"}, {\"evidence\": [{\"bytes\": 1999, \"path\": \"evidence/0-preflight.json\", \"sha256\": \"ac5667e646ae4bbf4af9d6a440fc2cb33d02737e8763f824adcbaf80b8f5e097\"}], \"name\": \"preflight\", \"sequence\": 1, \"status\": \"completed\"}, {\"evidence\": [{\"bytes\": 34508, \"path\": \"evidence/0-context.json\", \"sha256\": \"fdbf1c12319247b08f6c2bb06f6948e2a1758e39c76806fcbe3de42d6c175671\"}], \"name\": \"context\", \"sequence\": 2, \"status\": \"completed\"}], \"summary\": {\"advertised_context\": 327680, \"attempted_buckets\": [32768, 257000], \"buckets\": [{\"completed_count\": 6, \"completion_rate\": 1.0, \"engine_telemetry\": {\"available\": false, \"observations\": []}, \"failures\": {}, \"latency_ms\": {\"available\": true, \"mean\": 7288.459410163341, \"observations\": [5695.6744770286605, 6167.059329978656, 9742.416748951655, 7444.750800030306, 8182.070677983575, 6498.784427007195]}, \"meets_policy\": true, \"pass_rate\": 1.0, \"passed_count\": 6, \"relative_drop_from_baseline\": 0.0, \"requested_tokens\": 32768, \"sample_count\": 6, \"samples\": [{\"answer_sha256\": \"165c06e1078f0d23d51237f55fdf201897c11ab1ba6b89b4be0058e8551485db\", \"case_id\": \"native-needle-32768-0.100-r0\", \"case_type\": \"native-needle\", \"completed\": true, \"distractor_count\": 0, \"engine_telemetry\": null, \"failure\": null, \"finish_reason\": \"stop\", \"latency_ms\": 5695.6744770286605, \"passed\": true, \"position\": 0.1, \"prompt_tokens\": 32658, \"raw_reasoning\": \"We need answer user's request. Need read reference doc and answer final question. User says \\\"Return only the access marker for ORCHID.\\\" In reference: \\\"The access marker for ORCHID is K557508437185.\\\" Need final only marker. Ensure no extra.\\n\", \"raw_reasoning_field\": \"reasoning\", \"raw_reasoning_truncated\": false, \"requested_tokens\": 32768, \"schema\": \"anvil-serving.context-observation/v1\", \"target_count\": 1, \"throughput_tps\": 14.221318357761268, \"token_measurement\": \"usage\", \"visible_answer\": \"\\n\\nK557508437185\", \"visible_answer_truncated\": false}, {\"answer_sha256\": \"165c06e1078f0d23d51237f55fdf201897c11ab1ba6b89b4be0058e8551485db\", \"case_id\": \"native-needle-32768-0.900-r0\", \"case_type\": \"native-needle\", \"completed\": true, \"distractor_count\": 0, \"engine_telemetry\": null, \"failure\": null, \"finish_reason\": \"stop\", \"latency_ms\": 6167.059329978656, \"passed\": true, \"position\": 0.9, \"prompt_tokens\": 32658, \"raw_reasoning\": \"We need answer user's request. Need analyze reference. User provided huge reference paragraphs neutral. Contains one sentence: \\\"The access marker for ORCHID is K557508437185.\\\" Question: Return only the access marker for ORCHID. Need final only K557508437185. Ensure no extra.\\n\", \"raw_reasoning_field\": \"reasoning\", \"raw_reasoning_truncated\": false, \"requested_tokens\": 32768, \"schema\": \"anvil-serving.context-observation/v1\", \"target_count\": 1, \"throughput_tps\": 15.40442452664531, \"token_measurement\": \"usage\", \"visible_answer\": \"\\n\\nK557508437185\", \"visible_answer_truncated\": false}, {\"answer_sha256\": \"ea77d2c40d5ae22b614345bf65faaf4a798b6926e2753eaa34d64261ffa8f4cb\", \"case_id\": \"native-order-32768-0.100-r0\", \"case_type\": \"native-order\", \"completed\": true, \"distractor_count\": 0, \"engine_telemetry\": null, \"failure\": null, \"finish_reason\": \"stop\", \"latency_ms\": 9742.416748951655, \"passed\": true, \"position\": 0.1, \"prompt_tokens\": 32662, \"raw_reasoning\": \"We need answer user's request. Need parse reference document. It asks: Return the ALPHA, BETA, and GAMMA checkpoint values in that order, separated by ' | '. Need find checkpoint lines:\\nCheckpoint ALPHA stores K557508437185.\\nCheckpoint BETA stores K434177192288.\\nCheckpoint GAMMA stores K503557252183.\\nNeed output exactly \\\"K557508437185 | K434177192288 | K503557252183\\\". Ensure no extra. Need final.\\n\", \"raw_reasoning_field\": \"reasoning\", \"raw_reasoning_truncated\": false, \"requested_tokens\": 32768, \"schema\": \"anvil-serving.context-observation/v1\", \"target_count\": 3, \"throughput_tps\": 20.118211430556062, \"token_measurement\": \"usage\", \"visible_answer\": \"\\n\\nK557508437185 | K434177192288 | K503557252183\", \"visible_answer_truncated\": false}, {\"answer_sha256\": \"ea77d2c40d5ae22b614345bf65faaf4a798b6926e2753eaa34d64261ffa8f4cb\", \"case_id\": \"native-order-32768-0.900-r0\", \"case_type\": \"native-order\", \"completed\": true, \"distractor_count\": 0, \"engine_telemetry\": null, \"failure\": null, \"finish_reason\": \"stop\", \"latency_ms\": 7444.750800030306, \"passed\": true, \"position\": 0.9, \"prompt_tokens\": 32662, \"raw_reasoning\": \"We need answer user's question. Need identify checkpoint ALPHA, BETA, GAMMA values. Reference: Checkpoint ALPHA stores K557508437185. BETA K434177192288. GAMMA K503557252183. Need return in order separated by ' | '. Ensure no extra. Need final answer.\\n\", \"raw_reasoning_field\": \"reasoning\", \"raw_reasoning_truncated\": false, \"requested_tokens\": 32768, \"schema\": \"anvil-serving.context-observation/v1\", \"target_count\": 3, \"throughput_tps\": 17.86493645958688, \"token_measurement\": \"usage\", \"visible_answer\": \"\\n\\nK557508437185 | K434177192288 | K503557252183\", \"visible_answer_truncated\": false}, {\"answer_sha256\": \"165c06e1078f0d23d51237f55fdf201897c11ab1ba6b89b4be0058e8551485db\", \"case_id\": \"native-distractor-32768-0.100-r0\", \"case_type\": \"native-distractor\", \"completed\": true, \"distractor_count\": 2, \"engine_telemetry\": null, \"failure\": null, \"finish_reason\": \"stop\", \"latency_ms\": 8182.070677983575, \"passed\": true, \"position\": 0.1, \"prompt_tokens\": 32668, \"raw_reasoning\": \"We need answer user's question. Need read reference doc. User provided many neutral paragraphs and three key lines:\\nArchive ORCHID-001 stores K434177192288.\\nArchive ORCHID-002 stores K503557252183.\\nThe canonical ORCHID marker is K557508437185.\\nQuestion: Return only the canonical ORCHID marker, ignoring numbered archives.\\nSo answer exactly K557508437185. Need final only that. Ensure no extra.\\n\", \"raw_reasoning_field\": \"reasoning\", \"raw_reasoning_truncated\": false, \"requested_tokens\": 32768, \"schema\": \"anvil-serving.context-observation/v1\", \"target_count\": 1, \"throughput_tps\": 18.57720447331305, \"token_measurement\": \"usage\", \"visible_answer\": \"\\n\\nK557508437185\", \"visible_answer_truncated\": false}, {\"answer_sha256\": \"165c06e1078f0d23d51237f55fdf201897c11ab1ba6b89b4be0058e8551485db\", \"case_id\": \"native-distractor-32768-0.900-r0\", \"case_type\": \"native-distractor\", \"completed\": true, \"distractor_count\": 2, \"engine_telemetry\": null, \"failure\": null, \"finish_reason\": \"stop\", \"latency_ms\": 6498.784427007195, \"passed\": true, \"position\": 0.9, \"prompt_tokens\": 32668, \"raw_reasoning\": \"We need answer user's request. Need parse: They provide reference paragraphs, archive lines:\\nArchive ORCHID-001 stores K434177192288.\\nArchive ORCHID-002 stores K503557252183.\\nThe canonical ORCHID marker is K557508437185.\\nQuestion: Return only the canonical ORCHID marker, ignoring numbered archives.\\nNeed final only canonical marker: K557508437185. Ensure no extra text. Final: K557508437185.\\n\", \"raw_reasoning_field\": \"reasoning\", \"raw_reasoning_truncated\": false, \"requested_tokens\": 32768, \"schema\": \"anvil-serving.context-observation/v1\", \"target_count\": 1, \"throughput_tps\": 25.08161362032813, \"token_measurement\": \"usage\", \"visible_answer\": \"\\n\\nK557508437185\", \"visible_answer_truncated\": false}], \"throughput_tps\": {\"available\": true, \"mean\": 18.54461814469845, \"observations\": [14.221318357761268, 15.40442452664531, 20.118211430556062, 17.86493645958688, 18.57720447331305, 25.08161362032813]}}, {\"completed_count\": 6, \"completion_rate\": 1.0, \"engine_telemetry\": {\"available\": false, \"observations\": []}, \"failures\": {}, \"latency_ms\": {\"available\": true, \"mean\": 26827.640500492027, \"observations\": [26228.81304199109, 23683.442378009204, 29453.74174497556, 28545.917531009763, 26757.583044993225, 26296.34526197333]}, \"meets_policy\": true, \"pass_rate\": 1.0, \"passed_count\": 6, \"relative_drop_from_baseline\": 0.0, \"requested_tokens\": 257000, \"sample_count\": 6, \"samples\": [{\"answer_sha256\": \"165c06e1078f0d23d51237f55fdf201897c11ab1ba6b89b4be0058e8551485db\", \"case_id\": \"native-needle-257000-0.100-r0\", \"case_type\": \"native-needle\", \"completed\": true, \"distractor_count\": 0, \"engine_telemetry\": null, \"failure\": null, \"finish_reason\": \"stop\", \"latency_ms\": 26228.81304199109, \"passed\": true, \"position\": 0.1, \"prompt_tokens\": 256878, \"raw_reasoning\": \"We need answer user's request: \\\"Read the reference document and answer the final question. ... Question: Return only the access marker for ORCHID.\\\" We need extract access marker from reference document. There is a line: \\\"The access marker for ORCHID is K557508437185.\\\" Need return only that marker. Need ensure final only K557508437185. No extra.\\n\", \"raw_reasoning_field\": \"reasoning\", \"raw_reasoning_truncated\": false, \"requested_tokens\": 257000, \"schema\": \"anvil-serving.context-observation/v1\", \"target_count\": 1, \"throughput_tps\": 4.270113169844678, \"token_measurement\": \"usage\", \"visible_answer\": \"\\n\\nK557508437185\", \"visible_answer_truncated\": false}, {\"answer_sha256\": \"165c06e1078f0d23d51237f55fdf201897c11ab1ba6b89b4be0058e8551485db\", \"case_id\": \"native-needle-257000-0.900-r0\", \"case_type\": \"native-needle\", \"completed\": true, \"distractor_count\": 0, \"engine_telemetry\": null, \"failure\": null, \"finish_reason\": \"stop\", \"latency_ms\": 23683.442378009204, \"passed\": true, \"position\": 0.9, \"prompt_tokens\": 256878, \"raw_reasoning\": \"We need answer user's request: \\\"Read the reference document and answer the final question. ... Question: Return only the access marker for ORCHID.\\\" Need output only marker. Need find in reference: \\\"The access marker for ORCHID is K557508437185.\\\" Ensure final only K557508437185.\\n\", \"raw_reasoning_field\": \"reasoning\", \"raw_reasoning_truncated\": false, \"requested_tokens\": 257000, \"schema\": \"anvil-serving.context-observation/v1\", \"target_count\": 1, \"throughput_tps\": 4.137911982381243, \"token_measurement\": \"usage\", \"visible_answer\": \"\\n\\nK557508437185\", \"visible_answer_truncated\": false}, {\"answer_sha256\": \"ea77d2c40d5ae22b614345bf65faaf4a798b6926e2753eaa34d64261ffa8f4cb\", \"case_id\": \"native-order-257000-0.100-r0\", \"case_type\": \"native-order\", \"completed\": true, \"distractor_count\": 0, \"engine_telemetry\": null, \"failure\": null, \"finish_reason\": \"stop\", \"latency_ms\": 29453.74174497556, \"passed\": true, \"position\": 0.1, \"prompt_tokens\": 256902, \"raw_reasoning\": \"We need answer user's request. Need extract checkpoint values from reference. They ask: Return ALPHA, BETA, GAMMA in that order separated by ' | '. Need final answer concise. Values:\\nCheckpoint ALPHA stores K557508437185.\\nCheckpoint BETA stores K434177192288.\\nCheckpoint GAMMA stores K503557252183.\\nFinal: K557508437185 | K434177192288 | K503557252183\\n\", \"raw_reasoning_field\": \"reasoning\", \"raw_reasoning_truncated\": false, \"requested_tokens\": 257000, \"schema\": \"anvil-serving.context-observation/v1\", \"target_count\": 3, \"throughput_tps\": 6.281035584606451, \"token_measurement\": \"usage\", \"visible_answer\": \"\\n\\nK557508437185 | K434177192288 | K503557252183\", \"visible_answer_truncated\": false}, {\"answer_sha256\": \"ea77d2c40d5ae22b614345bf65faaf4a798b6926e2753eaa34d64261ffa8f4cb\", \"case_id\": \"native-order-257000-0.900-r0\", \"case_type\": \"native-order\", \"completed\": true, \"distractor_count\": 0, \"engine_telemetry\": null, \"failure\": null, \"finish_reason\": \"stop\", \"latency_ms\": 28545.917531009763, \"passed\": true, \"position\": 0.9, \"prompt_tokens\": 256902, \"raw_reasoning\": \"We need answer user's request: \\\"Read the reference document and answer the final question. ... Question: Return the ALPHA, BETA, and GAMMA checkpoint values in that order, separated by ' | '.\\\"\\n\\nNeed find checkpoint values in provided text. It includes many reference paragraphs, then three lines:\\nCheckpoint ALPHA stores K557508437185.\\nCheckpoint BETA stores K434177192288.\\nCheckpoint GAMMA stores K503557252183.\\n\\nNeed output exactly \\\"K557508437185 | K434177192288 | K503557252183\\\". Ensure separated by ' | '. No extra.\\n\", \"raw_reasoning_field\": \"reasoning\", \"raw_reasoning_truncated\": false, \"requested_tokens\": 257000, \"schema\": \"anvil-serving.context-observation/v1\", \"target_count\": 3, \"throughput_tps\": 7.6718500907213, \"token_measurement\": \"usage\", \"visible_answer\": \"\\n\\nK557508437185 | K434177192288 | K503557252183\", \"visible_answer_truncated\": false}, {\"answer_sha256\": \"165c06e1078f0d23d51237f55fdf201897c11ab1ba6b89b4be0058e8551485db\", \"case_id\": \"native-distractor-257000-0.100-r0\", \"case_type\": \"native-distractor\", \"completed\": true, \"distractor_count\": 2, \"engine_telemetry\": null, \"failure\": null, \"finish_reason\": \"stop\", \"latency_ms\": 26757.583044993225, \"passed\": true, \"position\": 0.1, \"prompt_tokens\": 256908, \"raw_reasoning\": \"We need answer user's request. Need read reference doc and answer final question. The document contains many neutral paragraphs, plus:\\nReference paragraph 0001283? Archive ORCHID-001 stores K434177192288.\\nArchive ORCHID-002 stores K503557252183.\\nThe canonical ORCHID marker is K557508437185.\\nQuestion: Return only the canonical ORCHID marker, ignoring numbered archives.\\nSo answer only canonical marker: K557508437185. Need no extra. Ensure final only marker.\\n\", \"raw_reasoning_field\": \"reasoning\", \"raw_reasoning_truncated\": false, \"requested_tokens\": 257000, \"schema\": \"anvil-serving.context-observation/v1\", \"target_count\": 1, \"throughput_tps\": 6.315966569769186, \"token_measurement\": \"usage\", \"visible_answer\": \"\\n\\nK557508437185\", \"visible_answer_truncated\": false}, {\"answer_sha256\": \"165c06e1078f0d23d51237f55fdf201897c11ab1ba6b89b4be0058e8551485db\", \"case_id\": \"native-distractor-257000-0.900-r0\", \"case_type\": \"native-distractor\", \"completed\": true, \"distractor_count\": 2, \"engine_telemetry\": null, \"failure\": null, \"finish_reason\": \"stop\", \"latency_ms\": 26296.34526197333, \"passed\": true, \"position\": 0.9, \"prompt_tokens\": 256908, \"raw_reasoning\": \"We need answer user's final request: \\\"Return only the canonical ORCHID marker, ignoring numbered archives.\\\" We have reference doc with many neutral paragraphs. Amid them:\\n\\\"Archive ORCHID-001 stores K434177192288.\\nArchive ORCHID-002 stores K503557252183.\\nThe canonical ORCHID marker is K557508437185.\\\"\\nQuestion asks return only canonical marker, ignoring numbered archives. So answer K557508437185. Need final only that.\\n\", \"raw_reasoning_field\": \"reasoning\", \"raw_reasoning_truncated\": false, \"requested_tokens\": 257000, \"schema\": \"anvil-serving.context-observation/v1\", \"target_count\": 1, \"throughput_tps\": 5.8943552214513515, \"token_measurement\": \"usage\", \"visible_answer\": \"\\n\\nK557508437185\", \"visible_answer_truncated\": false}], \"throughput_tps\": {\"available\": true, \"mean\": 5.761872103129035, \"observations\": [4.270113169844678, 4.137911982381243, 6.281035584606451, 7.6718500907213, 6.315966569769186, 5.8943552214513515]}}], \"effective_context\": 257000, \"first_material_degradation\": null, \"notes\": [\"effective_context uses attempted buckets only\", \"missing engine telemetry is unavailable and is not inferred from latency\"], \"schema\": \"anvil-serving.context-curve/v1\", \"threshold_policy\": {\"baseline_bucket\": 32768, \"baseline_pass_rate\": 1.0, \"max_relative_drop\": 0.25, \"pass_rate_floor\": 0.75}}}}, \"run\": {\"ownership_id\": \"swift-tp2-20260926\", \"profile\": \"scout\", \"run_id\": \"swift-tp2-r3-context\", \"spec_sha256\": \"8fba4c78bd0a714f3df9c0a35caf09c4daa9ef015bebef3442b6bd16d085ab57\", \"suite\": \"context\"}, \"schema\": \"anvil-serving.benchmark-result/v1\", \"status\": \"completed\"}, \"ok\": true}\n","error":null,"ok":true,"warnings":[]}