{ "context": { "cap_tokens": 1039360, "max_model_len": 1048576, "targets": [ { "attempted_context_tokens": 4096, "chars_per_token": 3.0, "clamped_tokens": 4096, "content_chunks": 10, "decode_tok_s": 80.1675632065896, "e2e_ms": 1604.2716000010842, "effective_prefill_tok_s": 2598.145174627506, "estimated_prompt_tokens": 4096, "generation_ms": 461.53330000379356, "mean_inter_token_latency_ms": 12.473872973075501, "output_token_source": "usage", "output_tokens": 38, "prompt_tokens": 2969, "reasoning_chunks": 3, "status": "passed", "target_tokens": 4096, "time_to_first_output_ms": 1142.7382999972906, "ttft_ms": 1249.680900000385, "usage": { "completion_tokens": 38, "prompt_tokens": 2969, "total_tokens": 3007 }, "visible_generation_ms": 354.59070000069914 }, { "attempted_context_tokens": 131072, "chars_per_token": 4.139, "clamped_tokens": 131072, "content_chunks": 14, "decode_tok_s": 70.47046256930876, "e2e_ms": 44464.55990000686, "effective_prefill_tok_s": 3311.4832029019926, "estimated_prompt_tokens": 131071, "generation_ms": 567.6137000045856, "mean_inter_token_latency_ms": 14.190342500114639, "output_token_source": "usage", "output_tokens": 41, "prompt_tokens": 145364, "reasoning_chunks": 4, "status": "passed", "target_tokens": 131072, "time_to_first_output_ms": 43896.946200002276, "ttft_ms": 43969.53900000517, "usage": { "completion_tokens": 41, "prompt_tokens": 145364, "total_tokens": 145405 }, "visible_generation_ms": 495.0209000016912 }, { "attempted_context_tokens": 240000, "chars_per_token": 3.732, "clamped_tokens": 240000, "content_chunks": 14, "decode_tok_s": 77.49608266746766, "e2e_ms": 75773.94139999524, "effective_prefill_tok_s": 3303.5070568469264, "estimated_prompt_tokens": 239999, "generation_ms": 529.0590000004158, "mean_inter_token_latency_ms": 12.90387804879063, "output_token_source": "usage", "output_tokens": 42, "prompt_tokens": 248572, "reasoning_chunks": 4, "status": "passed", "target_tokens": 240000, "time_to_first_output_ms": 75244.88239999482, "ttft_ms": 75313.54879999708, "usage": { "completion_tokens": 42, "prompt_tokens": 248572, "total_tokens": 248614 }, "visible_generation_ms": 460.3925999981584 } ] }, "evaluation_protocol": { "budget_semantics": "visible_answer_tokens plus reasoning_headroom_tokens are sent as one max_tokens cap; the endpoint does not hard-partition the channels", "minimum_comparable_repetitions": 3, "minimum_pass_rate": 1.0, "reasoning_headroom_tokens": 7168, "records_finish_reason": true, "records_full_visible_answer": true, "records_reasoning_channel_metadata": true, "repetitions": 3, "version": 3, "visible_answer_tokens": 1024 }, "failures": [], "identity": { "base_url": "http://127.0.0.1:8001/v1", "candidate_id": "wrldsuksgo2mars-glm53-exl3-k3-319d66a", "config_id": "dflash2-k5-fp8-1m-vision-wsl2", "engine": "vllm-glm53-exl3-k3-b12x-dflash2-k5", "gpu": "dual-rtx-pro-6000-max-q-tp2", "model": "glm53-flash-exl3-k3-dflash2-k5-fp8-tp2-1m-vision", "started_at": "2026-08-29T23:47:29Z" }, "intelligence": { "checks": [ { "attempt_count": 3, "attempts": [ { "attempt": 1, "budget": { "legacy_max_tokens_as_visible": false, "max_completion_tokens": 8192, "reasoning_headroom_tokens": 7168, "visible_answer_tokens": 1024 }, "content": "```diff\n--- app.py\n+++ app.py\n@@ -1,3 +1,3 @@\n-timeout = 30\n+timeout = 45\n retries = 2\n```", "content_chars": 90, "content_excerpt": "```diff\n--- app.py\n+++ app.py\n@@ -1,3 +1,3 @@\n-timeout = 30\n+timeout = 45\n retries = 2\n```", "failure_class": null, "finish_reason": "stop", "latency_ms": 542.5667000017711, "reasoning_chars": 12, "reasoning_excerpt": "Simple diff.", "reasoning_field": "reasoning", "reasoning_tokens": null, "status": "passed", "text_checks": [ { "name": "diff_shape", "passed": true }, { "name": "removes_old_timeout", "passed": true }, { "name": "adds_new_timeout", "passed": true } ], "usage": { "completion_tokens": 44, "prompt_tokens": 48, "prompt_tokens_details": null, "total_tokens": 92 } }, { "attempt": 2, "budget": { "legacy_max_tokens_as_visible": false, "max_completion_tokens": 8192, "reasoning_headroom_tokens": 7168, "visible_answer_tokens": 1024 }, "content": "```diff\n--- app.py\n+++ app.py\n@@ -1,3 +1,3 @@\n-timeout = 30\n+timeout = 45\n retries = 2\n```", "content_chars": 90, "content_excerpt": "```diff\n--- app.py\n+++ app.py\n@@ -1,3 +1,3 @@\n-timeout = 30\n+timeout = 45\n retries = 2\n```", "failure_class": null, "finish_reason": "stop", "latency_ms": 507.95699999434873, "reasoning_chars": 12, "reasoning_excerpt": "Simple diff.", "reasoning_field": "reasoning", "reasoning_tokens": null, "status": "passed", "text_checks": [ { "name": "diff_shape", "passed": true }, { "name": "removes_old_timeout", "passed": true }, { "name": "adds_new_timeout", "passed": true } ], "usage": { "completion_tokens": 44, "prompt_tokens": 48, "prompt_tokens_details": null, "total_tokens": 92 } }, { "attempt": 3, "budget": { "legacy_max_tokens_as_visible": false, "max_completion_tokens": 8192, "reasoning_headroom_tokens": 7168, "visible_answer_tokens": 1024 }, "content": "```diff\n--- app.py\n+++ app.py\n@@ -1,3 +1,3 @@\n-timeout = 30\n+timeout = 45\n retries = 2\n```", "content_chars": 90, "content_excerpt": "```diff\n--- app.py\n+++ app.py\n@@ -1,3 +1,3 @@\n-timeout = 30\n+timeout = 45\n retries = 2\n```", "failure_class": null, "finish_reason": "stop", "latency_ms": 547.3405000011553, "reasoning_chars": 12, "reasoning_excerpt": "Simple diff.", "reasoning_field": "reasoning", "reasoning_tokens": null, "status": "passed", "text_checks": [ { "name": "diff_shape", "passed": true }, { "name": "removes_old_timeout", "passed": true }, { "name": "adds_new_timeout", "passed": true } ], "usage": { "completion_tokens": 44, "prompt_tokens": 48, "prompt_tokens_details": null, "total_tokens": 92 } } ], "id": "unified_diff_timeout_edit", "pass_count": 3, "pass_rate": 1.0, "required_pass_rate": 1.0, "status": "passed", "validator": "deterministic_text_checks" }, { "attempt_count": 3, "attempts": [ { "attempt": 1, "budget": { "legacy_max_tokens_as_visible": false, "max_completion_tokens": 8192, "reasoning_headroom_tokens": 7168, "visible_answer_tokens": 1024 }, "content": "The pipeline's total latency (550+1800+650 = 3000 ms) exceeds the 2500 ms budget by 500 ms, driven mainly by the LLM; a practical fix is to stream LLM tokens directly into TTS so first audio can start before generation completes (or use a faster/smaller model).", "content_chars": 261, "content_excerpt": "The pipeline's total latency (550+1800+650 = 3000 ms) exceeds the 2500 ms budget by 500 ms, driven mainly by the LLM; a practical fix is to stream LLM tokens directly into TTS so first audio can start", "failure_class": null, "finish_reason": "stop", "latency_ms": 1518.1308000028366, "reasoning_chars": 166, "reasoning_excerpt": "STT+LLM+TTS = 3000 ms > 2500 ms budget. Problem: pipeline exceeds timeout by 500 ms; fix: stream LLM tokens to TTS (streaming/pipelining) or use faster/smaller model.", "reasoning_field": "reasoning", "reasoning_tokens": null, "status": "passed", "text_checks": [ { "name": "identifies_budget_overrun", "passed": true }, { "name": "offers_latency_fix", "passed": true } ], "usage": { "completion_tokens": 122, "prompt_tokens": 71, "prompt_tokens_details": null, "total_tokens": 193 } }, { "attempt": 2, "budget": { "legacy_max_tokens_as_visible": false, "max_completion_tokens": 8192, "reasoning_headroom_tokens": 7168, "visible_answer_tokens": 1024 }, "content": "The pipeline's total latency (550+1800+650 = 3000 ms) exceeds the 2500 ms budget by 500 ms, driven mainly by the LLM; a practical fix is to stream LLM tokens directly into TTS so first audio can start before generation completes (or use a faster/smaller model).", "content_chars": 261, "content_excerpt": "The pipeline's total latency (550+1800+650 = 3000 ms) exceeds the 2500 ms budget by 500 ms, driven mainly by the LLM; a practical fix is to stream LLM tokens directly into TTS so first audio can start", "failure_class": null, "finish_reason": "stop", "latency_ms": 1555.426000006264, "reasoning_chars": 166, "reasoning_excerpt": "STT+LLM+TTS = 3000 ms > 2500 ms budget. Problem: pipeline exceeds timeout by 500 ms; fix: stream LLM tokens to TTS (streaming/pipelining) or use faster/smaller model.", "reasoning_field": "reasoning", "reasoning_tokens": null, "status": "passed", "text_checks": [ { "name": "identifies_budget_overrun", "passed": true }, { "name": "offers_latency_fix", "passed": true } ], "usage": { "completion_tokens": 122, "prompt_tokens": 71, "prompt_tokens_details": null, "total_tokens": 193 } }, { "attempt": 3, "budget": { "legacy_max_tokens_as_visible": false, "max_completion_tokens": 8192, "reasoning_headroom_tokens": 7168, "visible_answer_tokens": 1024 }, "content": "The pipeline's total latency (550+1800+650 = 3000 ms) exceeds the 2500 ms budget by 500 ms, driven mainly by the LLM; a practical fix is to stream LLM tokens directly into TTS so audio playback starts before generation completes (or use a faster/smaller model).", "content_chars": 261, "content_excerpt": "The pipeline's total latency (550+1800+650 = 3000 ms) exceeds the 2500 ms budget by 500 ms, driven mainly by the LLM; a practical fix is to stream LLM tokens directly into TTS so audio playback starts", "failure_class": null, "finish_reason": "stop", "latency_ms": 1531.5496000039275, "reasoning_chars": 175, "reasoning_excerpt": "STT+LLM+TTS = 3000 ms > 2500 ms budget. Problem: pipeline exceeds timeout by 500 ms; fix: stream LLM tokens to TTS (or reduce LLM latency via smaller/faster model, streaming).", "reasoning_field": "reasoning", "reasoning_tokens": null, "status": "passed", "text_checks": [ { "name": "identifies_budget_overrun", "passed": true }, { "name": "offers_latency_fix", "passed": true } ], "usage": { "completion_tokens": 121, "prompt_tokens": 71, "prompt_tokens_details": null, "total_tokens": 192 } } ], "id": "parallel_timeout_triage", "pass_count": 3, "pass_rate": 1.0, "required_pass_rate": 1.0, "status": "passed", "validator": "deterministic_text_checks" } ], "status": "passed" }, "run_id": "fast-bakeoff-20260829T234729Z", "schema": "anvil-serving.fast-tier-bakeoff/v1", "score_inputs": { "e2e_p50_ms": 44464.55990000686, "intelligence_pass_rate": 1.0, "operational_fit_notes": [ "endpoint was already loaded; benchmark did not start or stop serves" ], "session_recall_passed": true, "thinking_mode": "default", "tool_call_passed": true, "ttft_p50_ms": 43969.53900000517, "usable_context_tokens": 240000, "voice_latency_ms": null }, "selection": { "context_targets": [ 4096, 131072, 240000 ], "endpoint_already_loaded": true, "requests_per_context": 1, "suites": [ "chat", "context", "tool", "session", "intelligence" ] }, "session": { "checks": [ { "attempt_count": 3, "attempts": [ { "attempt": 1, "budget": { "legacy_max_tokens_as_visible": false, "max_completion_tokens": 8192, "reasoning_headroom_tokens": 7168, "visible_answer_tokens": 1024 }, "content": "RIVER-918", "content_chars": 9, "content_excerpt": "RIVER-918", "expected": "RIVER-918", "failure_class": null, "finish_reason": "stop", "latency_ms": 360.4890999995405, "reasoning_chars": 88, "reasoning_excerpt": "The user asked me to remember \"RIVER-918\" and now asks what session code should be used.", "reasoning_field": "reasoning", "reasoning_tokens": null, "status": "passed", "usage": { "completion_tokens": 30, "prompt_tokens": 45, "prompt_tokens_details": null, "total_tokens": 75 } }, { "attempt": 2, "budget": { "legacy_max_tokens_as_visible": false, "max_completion_tokens": 8192, "reasoning_headroom_tokens": 7168, "visible_answer_tokens": 1024 }, "content": "RIVER-918", "content_chars": 9, "content_excerpt": "RIVER-918", "expected": "RIVER-918", "failure_class": null, "finish_reason": "stop", "latency_ms": 384.3789000020479, "reasoning_chars": 88, "reasoning_excerpt": "The user asked me to remember \"RIVER-918\" and now asks what session code should be used.", "reasoning_field": "reasoning", "reasoning_tokens": null, "status": "passed", "usage": { "completion_tokens": 30, "prompt_tokens": 45, "prompt_tokens_details": null, "total_tokens": 75 } }, { "attempt": 3, "budget": { "legacy_max_tokens_as_visible": false, "max_completion_tokens": 8192, "reasoning_headroom_tokens": 7168, "visible_answer_tokens": 1024 }, "content": "RIVER-918", "content_chars": 9, "content_excerpt": "RIVER-918", "expected": "RIVER-918", "failure_class": null, "finish_reason": "stop", "latency_ms": 365.16000000119675, "reasoning_chars": 88, "reasoning_excerpt": "The user asked me to remember \"RIVER-918\" and now asks what session code should be used.", "reasoning_field": "reasoning", "reasoning_tokens": null, "status": "passed", "usage": { "completion_tokens": 30, "prompt_tokens": 45, "prompt_tokens_details": null, "total_tokens": 75 } } ], "name": "single_request_multiturn_recall", "pass_count": 3, "pass_rate": 1.0, "required_pass_rate": 1.0, "status": "passed" } ], "status": "passed" }, "source_recipe": { "ref": "configs/glm53-flash-purtell-k3-dflash2-fp8-1m-vision-sm120-tp2-wsl2-recipe.toml", "serve_command": null }, "suites": {}, "thinking": { "chat_template_kwargs": null, "control_evidence": null, "control_evidence_sha256": null, "control_mechanism": "reasoning_effort", "control_requested": "high", "control_status": "requested_unverified", "mode": "default", "reasoning_effort": "high", "unsupported": false }, "timing": { "chat": { "content_chunks": 38, "decode_tok_s_p50": 77.49608266746766, "decode_tok_s_p95": 80.1675632065896, "e2e_p50_ms": 44464.55990000686, "e2e_p95_ms": 75773.94139999524, "effective_prefill_tok_s_p50": 3303.5070568469264, "effective_prefill_tok_s_p95": 3311.4832029019926, "generation_p50_ms": 529.0590000004158, "generation_p95_ms": 567.6137000045856, "mean_inter_token_latency_ms_p50": 12.90387804879063, "mean_inter_token_latency_ms_p95": 14.190342500114639, "output_tokens": 121, "prompt_token_samples": 3, "prompt_tokens": 396905, "prompt_tokens_p50": 145364.0, "prompt_tokens_p95": 248572.0, "reasoning_chunks": 11, "time_to_first_output_p50_ms": 43896.946200002276, "time_to_first_output_p95_ms": 75244.88239999482, "ttft_p50_ms": 43969.53900000517, "ttft_p95_ms": 75313.54879999708, "ttft_samples": 3 }, "wall_ms": 130109.72039999615 }, "tool": { "checks": [ { "attempt_count": 3, "attempts": [ { "arguments": { "zip": "98101" }, "attempt": 1, "budget": { "legacy_max_tokens_as_visible": false, "max_completion_tokens": 8192, "reasoning_headroom_tokens": 7168, "visible_answer_tokens": 1024 }, "content": "", "content_chars": 0, "content_excerpt": "", "failure_class": null, "finish_reason": "tool_calls", "latency_ms": 354.0433000016492, "reasoning_chars": 0, "reasoning_excerpt": "", "reasoning_field": null, "reasoning_tokens": null, "status": "passed", "tool_call_count": 1, "usage": { "completion_tokens": 15, "prompt_tokens": 171, "prompt_tokens_details": null, "total_tokens": 186 }, "valid_tool_call_count": 1, "validation_errors": [] }, { "arguments": { "zip": "98101" }, "attempt": 2, "budget": { "legacy_max_tokens_as_visible": false, "max_completion_tokens": 8192, "reasoning_headroom_tokens": 7168, "visible_answer_tokens": 1024 }, "content": "", "content_chars": 0, "content_excerpt": "", "failure_class": null, "finish_reason": "tool_calls", "latency_ms": 301.22359999950277, "reasoning_chars": 0, "reasoning_excerpt": "", "reasoning_field": null, "reasoning_tokens": null, "status": "passed", "tool_call_count": 1, "usage": { "completion_tokens": 15, "prompt_tokens": 171, "prompt_tokens_details": null, "total_tokens": 186 }, "valid_tool_call_count": 1, "validation_errors": [] }, { "arguments": { "zip": "98101" }, "attempt": 3, "budget": { "legacy_max_tokens_as_visible": false, "max_completion_tokens": 8192, "reasoning_headroom_tokens": 7168, "visible_answer_tokens": 1024 }, "content": "", "content_chars": 0, "content_excerpt": "", "failure_class": null, "finish_reason": "tool_calls", "latency_ms": 287.95440000249073, "reasoning_chars": 0, "reasoning_excerpt": "", "reasoning_field": null, "reasoning_tokens": null, "status": "passed", "tool_call_count": 1, "usage": { "completion_tokens": 15, "prompt_tokens": 171, "prompt_tokens_details": null, "total_tokens": 186 }, "valid_tool_call_count": 1, "validation_errors": [] } ], "expected_arguments": { "zip": "98101" }, "expected_function": "record_weather_zip", "name": "openai_tool_call_smoke", "pass_count": 3, "pass_rate": 1.0, "required_pass_rate": 1.0, "status": "passed" } ], "status": "passed" }, "voice": { "status": "not_run", "stt_latency_ms": null, "total_turn_latency_ms": null, "tts_latency_ms": null } }