CL-WA-llm-latency-short-request-counts-and-endpoints
DESCRIPTIVE reproducible from public raw logs
- Metric
client_visible_latency- Analysis rule
Keep schedule capacity, started requests, terminal requests, uncensored completions, and censored/unstarted requests separate; quantiles use completed requests only.- Publication status
STATUS_ONLY- Evidence status
reproducible from public raw logs- Interpretation
Client-visible timing is not provider transport latency, task success, or a Kogen Build efficacy result.- Reported values
{"censored_n":1,"graded_n":0,"per_arm":{"gpt-6-luna/high":{"attempts":14,"complete":13,"first_stdout_p50_s":0.708,"first_stdout_p90_s":0.987,"wall_p50_s":5.331,"wall_p90_s":6.722},"gpt-6-luna/max":{"attempts":12,"complete":12,"first_stdout_p50_s":0.67,"first_stdout_p90_s":1.021,"wall_p50_s":7.116,"wall_p90_s":8.266},"gpt-6-luna/medium":{"attempts":15,"complete":15,"first_stdout_p50_s":0.723,"first_stdout_p90_s":1.363,"wall_p50_s":5.416,"wall_p90_s":7.256},"gpt-6-luna/xhigh":{"attempts":12,"complete":12,"first_stdout_p50_s":0.653,"first_stdout_p90_s":0.939,"wall_p50_s":6.82,"wall_p90_s":8.216},"gpt-6.1-sol/high":{"attempts":12,"complete":12,"first_stdout_p50_s":0.738,"first_stdout_p90_s":1.002,"wall_p50_s":7.342,"wall_p90_s":8.902},"gpt-6.1-sol/medium":{"attempts":12,"complete":12,"first_stdout_p50_s":0.922,"first_stdout_p90_s":1.149,"wall_p50_s":6.973,"wall_p90_s":7.958}},"request_attempts_n":77,"scheduled_capacity_n":36,"terminal_records_n":77,"uncensored_completed_n":76}
Round: llm-latency-short