Round score
-1.1913
weighted tier speedup after failure cap and deduction
Eligible requests
31
trusted baseline exclusions removed
Failed eligible requests
8
each request counted once
Raw weighted speedup
-116.55%
before failure cap and deduction
Raw weighted speedup
Eligible speedup
Request failures cap speed credit at zero: min(-1.165481, 0) = -1.165481. Negative speedups remain negative.
Failed requests
Deduction
Final score
Each duration is the median across repetitions, from the group start until the tier's last request finishes. This includes client queueing. Weighted speedup = sum of weight × (1 − candidate / baseline).
| Input tier | Eligible requests | Weight | Baseline | Candidate | Speedup | Contribution |
|---|
The round score uses weighted completion times for all eligible work in each tier. Per-request speedups below are diagnostics, not score contributions. Only trusted baseline exclusions are removed; candidate failures remain in the eligible count and incur the configured failure penalty.
31 prompts
| Prompt | Diagnostic speedup | Baseline | Candidate | Aligned output | Diagnostic / failure |
|---|---|---|---|---|---|
| hf-025 | — | — | — | 4224 | Per-token timing unavailable (not a request failure) |
| hf-016 | — | — | — | 3097 | Per-token timing unavailable (not a request failure) |
| hf-019 | — | — | — |
31 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 2.617s | 89.232s | 4625 | 23.529s | 1988 gaps for 4625 tokens |
| hf-001 | 4.880s | 80.854s | 2747 | 21.265s | 1828 gaps for 2747 tokens |
| hf-002 | 12.951s | 98.643s |
Occupied client slots include protocol completion and final drain. Mean occupancy is weighted by elapsed time; it can be below the requested cap.
| Repetition | Group start | Requested | Effective cap | Peak | Mean |
|---|---|---|---|---|---|
| 1 | 0.000s | C32 | 31 | 31 | 19.28 |
| 2 | 0.000s | C32 | 31 | 31 | 19.24 |
| 3 | 0.000s | C32 | 31 | 31 | 19.59 |
Each latency-median response is compared with the lowest valid opening-baseline ratio for the same prompt. Drops above the recorded limit fail; absolute checks also apply. These checks detect repetition, not overall writing quality.
| Prompt | Candidate char-16 |
|---|
| 16k | 7 | 25% | 114.051s | 230.704s | -102.28% | -25.57% |
|---|
| 32k | 8 | 25% | 115.376s | 250.629s | -117.23% | -29.31% |
|---|
| 64k | 8 | 25% | 114.916s | 252.654s | -119.86% | -29.96% |
|---|
| 128k | 8 | 25% | 112.531s | 255.247s | -126.82% | -31.71% |
|---|
| 3913 |
| Per-token timing unavailable (not a request failure) |
| hf-001 | — | — | — | 2747 | candidate output below tolerance |
| hf-009 | — | — | — | 5046 | Per-token timing unavailable (not a request failure) |
| hf-006 | — | — | — | 4203 | candidate output below tolerance |
| hf-020 | — | — | — | 3101 | Per-token timing unavailable (not a request failure) |
| hf-000 | — | — | — | 4458 | Per-token timing unavailable (not a request failure) |
| hf-017 | — | — | — | 3989 | candidate output below tolerance |
| hf-012 | — | — | — | 3796 | candidate output below tolerance |
| hf-021 | — | — | — | 3778 | Per-token timing unavailable (not a request failure) |
| hf-023 | — | — | — | 3592 | candidate output below tolerance |
| hf-029 | — | — | — | 4259 | Per-token timing unavailable (not a request failure) |
| hf-022 | — | — | — | 3523 | Per-token timing unavailable (not a request failure) |
| hf-027 | — | — | — | 5120 | Per-token timing unavailable (not a request failure) |
| hf-018 | — | — | — | 4647 | Per-token timing unavailable (not a request failure) |
| hf-024 | — | — | — | 5120 | Per-token timing unavailable (not a request failure) |
| hf-004 | — | — | — | 3062 | Per-token timing unavailable (not a request failure) |
| hf-002 | — | — | — | 3862 | Per-token timing unavailable (not a request failure) |
| hf-026 | — | — | — | 5019 | Per-token timing unavailable (not a request failure) |
| hf-005 | — | — | — | 5120 | Per-token timing unavailable (not a request failure) |
| hf-007 | — | — | — | 5120 | Per-token timing unavailable (not a request failure) |
| hf-031 | — | — | — | 5120 | Per-token timing unavailable (not a request failure) |
| hf-015 | — | — | — | 5120 | Per-token timing unavailable (not a request failure) |
| hf-013 | — | — | — | 5120 | Per-token timing unavailable (not a request failure) |
| hf-030 | — | — | — | 5120 | Per-token timing unavailable (not a request failure) |
| hf-028 | — | — | — | 5120 | Per-token timing unavailable (not a request failure) |
| hf-011 | — | — | — | 4420 | candidate output below tolerance |
| hf-003 | — | — | — | 3968 | Per-token timing unavailable (not a request failure) |
| hf-014 | — | — | — | 3550 | candidate output below tolerance |
| hf-010 | — | — | — | 3658 | candidate output below tolerance |
| 3862 |
| 13.060s |
| 2235 gaps for 3862 tokens |
| hf-003 | 1.179s | 104.896s | 5120 | 20.842s | 2861 gaps for 5120 tokens |
| hf-004 | 11.499s | 79.385s | 4091 | 10.518s | 2172 gaps for 4091 tokens |
| hf-005 | 21.747s | 99.775s | 5120 | 6.539s | 2280 gaps for 5120 tokens |
| hf-006 | 21.805s | 65.446s | 4203 | 2.716s | 1806 gaps for 4203 tokens |
| hf-007 | 25.938s | 109.958s | 5120 | 6.539s | 2506 gaps for 5120 tokens |
| hf-009 | 23.183s | 90.729s | 5046 | 6.539s | 2113 gaps for 5046 tokens |
| hf-010 | 5.629s | 96.451s | 3658 | 20.382s | 2203 gaps for 3658 tokens |
| hf-011 | 82.983s | 176.393s | 4420 | 10.573s | 2680 gaps for 4420 tokens |
| hf-012 | 132.666s | 186.533s | 3796 | 10.573s | 1715 gaps for 3796 tokens |
| hf-013 | 17.328s | 117.383s | 5120 | 8.811s | 2707 gaps for 5120 tokens |
| hf-014 | 76.706s | 147.743s | 3550 | 5.721s | 2208 gaps for 3550 tokens |
| hf-015 | 71.351s | 150.907s | 5120 | 5.721s | 2508 gaps for 5120 tokens |
| hf-016 | 26.113s | 85.005s | 4032 | 5.904s | 1877 gaps for 4032 tokens |
| hf-017 | 97.862s | 143.391s | 3989 | 5.712s | 1444 gaps for 3989 tokens |
| hf-018 | 97.070s | 173.930s | 4647 | 7.635s | 2400 gaps for 4647 tokens |
| hf-019 | 149.585s | 223.309s | 5120 | 10.573s | 2195 gaps for 5120 tokens |
| hf-020 | 130.396s | 202.095s | 3356 | 10.573s | 2102 gaps for 3356 tokens |
| hf-021 | 133.480s | 199.432s | 3819 | 10.573s | 1978 gaps for 3819 tokens |
| hf-022 | 83.834s | 160.260s | 4257 | 9.891s | 2430 gaps for 4257 tokens |
| hf-023 | 164.552s | 217.524s | 3592 | 6.128s | 1468 gaps for 3592 tokens |
| hf-024 | 177.488s | 239.333s | 5120 | 6.128s | 2277 gaps for 5120 tokens |
| hf-025 | 150.132s | 217.900s | 5120 | 8.048s | 1980 gaps for 5120 tokens |
| hf-026 | 104.546s | 193.738s | 5108 | 8.048s | 2520 gaps for 5108 tokens |
| hf-027 | 154.236s | 230.674s | 5120 | 7.635s | 2242 gaps for 5120 tokens |
| hf-028 | 193.402s | 250.593s | 5120 | 6.138s | 2674 gaps for 5120 tokens |
| hf-029 | 204.393s | 242.207s | 4259 | 6.138s | 1703 gaps for 4259 tokens |
| hf-030 | 192.913s | 249.853s | 5120 | 6.138s | 2615 gaps for 5120 tokens |
| hf-031 | 217.672s | 255.208s | 5120 | 25.65ms | 2423 gaps for 5120 tokens |
| Baseline char-16 |
|---|
| Drop |
|---|
| Limit |
|---|
| Result |
|---|
| hf-000 | 0.8419 | 0.8349 | -0.0070 | 0.1000 | Passed |
| hf-001 | 0.9698 | 0.9675 | -0.0024 | 0.1000 | Passed |
| hf-002 | 0.8727 | 0.8724 | -0.0003 | 0.1000 | Passed |
| hf-003 | 0.6129 | 0.6893 | 0.0765 | 0.1000 | Passed |
| hf-004 | 0.9685 | 0.8909 | -0.0776 | 0.1000 | Passed |
| hf-005 | 0.6825 | 0.7822 | 0.0997 | 0.1000 | Passed |
| hf-006 | 0.8615 | 0.5629 | -0.2987 | 0.1000 | Passed |
| hf-007 | 0.8141 | 0.7446 | -0.0695 | 0.1000 | Passed |
| hf-009 | 0.9136 | 0.8359 | -0.0777 | 0.1000 | Passed |
| hf-010 | 0.8775 | 0.7058 | -0.1717 | 0.1000 | Passed |
| hf-011 | 0.9626 | 0.8447 | -0.1179 | 0.1000 | Passed |
| hf-012 | 0.9467 | 0.7309 | -0.2158 | 0.1000 | Passed |
| hf-013 | 0.7830 | 0.8118 | 0.0288 | 0.1000 | Passed |
| hf-014 | 0.9656 | 0.7324 | -0.2331 | 0.1000 | Passed |
| hf-015 | 0.7915 | 0.8284 | 0.0368 | 0.1000 | Passed |
| hf-016 | 0.6848 | 0.6848 | -0.0000 | 0.1000 | Passed |
| hf-017 | 0.8554 | 0.8503 | -0.0051 | 0.1000 | Passed |
| hf-018 | 0.9584 | 0.7754 | -0.1830 | 0.1000 | Passed |
| hf-019 | 0.6565 | 0.7449 | 0.0884 | 0.1000 | Passed |
| hf-020 | 0.9066 | 0.9516 | 0.0450 | 0.1000 | Passed |
| hf-021 | 0.8841 | 0.8648 | -0.0193 | 0.1000 | Passed |
| hf-022 | 0.8070 | 0.8203 | 0.0133 | 0.1000 | Passed |
| hf-023 | 0.8141 | 0.7202 | -0.0939 | 0.1000 | Passed |
| hf-024 | 0.7469 | 0.6893 | -0.0576 | 0.1000 | Passed |
| hf-025 | 0.8707 | 0.9327 | 0.0620 | 0.1000 | Passed |
| hf-026 | 0.9311 | 0.7746 | -0.1565 | 0.1000 | Passed |
| hf-027 | 0.7914 | 0.6934 | -0.0980 | 0.1000 | Passed |
| hf-028 | 0.6940 | 0.6553 | -0.0387 | 0.1000 | Passed |
| hf-029 | 0.9349 | 0.6717 | -0.2632 | 0.1000 | Passed |
| hf-030 | 0.8558 | 0.7808 | -0.0750 | 0.1000 | Passed |
| hf-031 | 0.7799 | 0.7712 | -0.0087 | 0.1000 | Passed |
Actual settings for each measured request. Each prompt keeps its temperature across repetitions and engines. Generation uses seed 0 throughout qualification, warmups and measured repetitions; fixed settings do not guarantee identical output text.
| Prompt | Repetition | Temperature | Top p | Seed |
|---|---|---|---|---|
| hf-002 | 1 | 0.308187 | 1 | 0 |
| hf-000 | 1 | 0.751507 | 1 | 0 |
| hf-006 | 1 | 0.806947 | 1 | 0 |
| hf-009 | 1 | 0.80922 | 1 | 0 |
| hf-004 | 1 | 0.705908 | 1 | 0 |
| hf-005 | 1 | 0.644584 | 1 | 0 |
| hf-003 | 1 | 0.165866 | 1 | 0 |
| hf-010 | 1 | 0.183241 | 1 | 0 |
| hf-001 | 1 | 0.145615 | 1 | 0 |
| hf-015 | 1 | 0.422296 | 1 | 0 |
| hf-016 | 1 | 0.477202 | 1 | 0 |
| hf-014 | 1 | 0.234039 | 1 | 0 |
| hf-017 | 1 | 0.948598 | 1 | 0 |
| hf-007 | 1 | 0.526161 | 1 | 0 |
| hf-027 | 1 | 0.616175 | 1 | 0 |
| hf-013 | 1 | 0.376028 | 1 | 0 |
| hf-022 | 1 | 0.135465 | 1 | 0 |
| hf-018 | 1 | 0.582993 | 1 | 0 |
| hf-011 | 1 | 0.229212 | 1 | 0 |
| hf-012 | 1 | 0.971562 | 1 | 0 |
| hf-021 | 1 | 0.394709 | 1 | 0 |
| hf-020 | 1 | 0.161701 | 1 | 0 |
| hf-026 | 1 | 0.634191 | 1 | 0 |
| hf-023 | 1 | 0.836006 | 1 | 0 |
| hf-025 | 1 | 0.985364 | 1 | 0 |
| hf-019 | 1 | 0.323977 | 1 | 0 |
| hf-024 | 1 | 0.606986 | 1 | 0 |
| hf-029 | 1 | 0.888902 | 1 | 0 |
| hf-030 | 1 | 0.466474 | 1 | 0 |
| hf-028 | 1 | 0.426835 | 1 | 0 |
| hf-031 | 1 | 0.466727 | 1 | 0 |
| hf-004 | 2 | 0.705908 | 1 | 0 |
| hf-010 | 2 | 0.183241 | 1 | 0 |
| hf-006 | 2 | 0.806947 | 1 | 0 |
| hf-001 | 2 | 0.145615 | 1 | 0 |
| hf-016 | 2 | 0.477202 | 1 | 0 |
| hf-000 | 2 | 0.751507 | 1 | 0 |
| hf-009 | 2 | 0.80922 | 1 | 0 |
| hf-005 | 2 | 0.644584 | 1 | 0 |
| hf-002 | 2 | 0.308187 | 1 | 0 |
| hf-007 | 2 | 0.526161 | 1 | 0 |
| hf-013 | 2 | 0.376028 | 1 | 0 |
| hf-003 | 2 | 0.165866 | 1 | 0 |
| hf-017 | 2 | 0.948598 | 1 | 0 |
| hf-022 | 2 | 0.135465 | 1 | 0 |
| hf-014 | 2 | 0.234039 | 1 | 0 |
| hf-015 | 2 | 0.422296 | 1 | 0 |
| hf-020 | 2 | 0.161701 | 1 | 0 |
| hf-018 | 2 | 0.582993 | 1 | 0 |
| hf-026 | 2 | 0.634191 | 1 | 0 |
| hf-011 | 2 | 0.229212 | 1 | 0 |
| hf-012 | 2 | 0.971562 | 1 | 0 |
| hf-025 | 2 | 0.985364 | 1 | 0 |
| hf-023 | 2 | 0.836006 | 1 | 0 |
| hf-029 | 2 | 0.888902 | 1 | 0 |
| hf-021 | 2 | 0.394709 | 1 | 0 |
| hf-019 | 2 | 0.323977 | 1 | 0 |
| hf-027 | 2 | 0.616175 | 1 | 0 |
| hf-024 | 2 | 0.606986 | 1 | 0 |
| hf-030 | 2 | 0.466474 | 1 | 0 |
| hf-028 | 2 | 0.426835 | 1 | 0 |
| hf-031 | 2 | 0.466727 | 1 | 0 |
| hf-006 | 3 | 0.806947 | 1 | 0 |
| hf-016 | 3 | 0.477202 | 1 | 0 |
| hf-001 | 3 | 0.145615 | 1 | 0 |
| hf-004 | 3 | 0.705908 | 1 | 0 |
| hf-009 | 3 | 0.80922 | 1 | 0 |
| hf-010 | 3 | 0.183241 | 1 | 0 |
| hf-002 | 3 | 0.308187 | 1 | 0 |
| hf-005 | 3 | 0.644584 | 1 | 0 |
| hf-003 | 3 | 0.165866 | 1 | 0 |
| hf-007 | 3 | 0.526161 | 1 | 0 |
| hf-013 | 3 | 0.376028 | 1 | 0 |
| hf-000 | 3 | 0.751507 | 1 | 0 |
| hf-017 | 3 | 0.948598 | 1 | 0 |
| hf-012 | 3 | 0.971562 | 1 | 0 |
| hf-014 | 3 | 0.234039 | 1 | 0 |
| hf-011 | 3 | 0.229212 | 1 | 0 |
| hf-022 | 3 | 0.135465 | 1 | 0 |
| hf-015 | 3 | 0.422296 | 1 | 0 |
| hf-018 | 3 | 0.582993 | 1 | 0 |
| hf-026 | 3 | 0.634191 | 1 | 0 |
| hf-021 | 3 | 0.394709 | 1 | 0 |
| hf-019 | 3 | 0.323977 | 1 | 0 |
| hf-025 | 3 | 0.985364 | 1 | 0 |
| hf-027 | 3 | 0.616175 | 1 | 0 |
| hf-023 | 3 | 0.836006 | 1 | 0 |
| hf-020 | 3 | 0.161701 | 1 | 0 |
| hf-024 | 3 | 0.606986 | 1 | 0 |
| hf-029 | 3 | 0.888902 | 1 | 0 |
| hf-030 | 3 | 0.466474 | 1 | 0 |
| hf-028 | 3 | 0.426835 | 1 | 0 |
| hf-031 | 3 | 0.466727 | 1 | 0 |
verdict
evidence
min logprob
num prompts
mean logprob
coverage ratio
quantile logprob
num positions scored
role
evidence
request concurrency
requests per s
sla goodput ratio
output tokens per s
p99 e2e ms rel range
p99 itl ms rel range
p99 ttft ms rel range
tier completion max rel range
Image digest
Patch hash