Round score
0.6532
weighted tier speedup after failure cap and deduction
Eligible requests
16
trusted baseline exclusions removed
Failed eligible requests
0
each request counted once
Raw weighted speedup
+65.32%
before failure cap and deduction
Temperature range
Request concurrency
Tier scheduling
Output policy
Thinking
Context limit
The same eligible inputs and output ceilings apply to every engine. Candidates must emit at least 90% of baseline tokens per request; natural output lengths can affect tier completion time. Slots refill until the group drains and stay occupied through protocol completion. Baseline exclusions and final drain can reduce actual concurrency below the configured cap.
Raw weighted speedup
Eligible speedup
No request failures: eligible speedup equals raw weighted speedup.
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.
16 prompts
| Prompt | Diagnostic speedup | Baseline | Candidate | Aligned output | Diagnostic / failure |
|---|---|---|---|---|---|
| hf-001 | — | — | — | 5120 | Per-token timing unavailable (not a request failure) |
| hf-006 | — | — | — | 4563 | Per-token timing unavailable (not a request failure) |
| hf-007 | — | — | — |
16 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 91.05ms | 22.291s | 4861 | 17.02ms | 1612 gaps for 4861 tokens |
| hf-001 | 97.08ms | 25.562s | 5120 | 45.92ms | 1833 gaps for 5120 tokens |
| hf-002 | 112.74ms | 27.857s |
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 | C4 | 4 | 4 | 3.55 |
| 1 | 49.370s | C4 | 4 | 4 | 3.81 |
| 2 | 0.000s | C4 | 4 | 4 | 3.91 |
| 2 | 48.239s | C4 | 4 | 4 | 3.84 |
| 3 | 0.000s | C4 | 4 | 4 | 3.70 |
| 3 | 49.845s | C4 | 4 | 4 | 3.86 |
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 |
|---|---|---|---|---|---|---|
| 8k | 8 | 50% | 148.925s | 49.368s | +66.85% | +33.43% |
| 16k | 8 | 50% | 147.801s | 53.529s | +63.78% | +31.89% |
| 5120 |
| Per-token timing unavailable (not a request failure) |
| hf-000 | — | — | — | 4861 | Per-token timing unavailable (not a request failure) |
| hf-011 | — | — | — | 4532 | Per-token timing unavailable (not a request failure) |
| hf-015 | — | — | — | 5081 | Per-token timing unavailable (not a request failure) |
| hf-012 | — | — | — | 4684 | Per-token timing unavailable (not a request failure) |
| hf-014 | — | — | — | 5120 | Per-token timing unavailable (not a request failure) |
| hf-005 | — | — | — | 4472 | Per-token timing unavailable (not a request failure) |
| hf-003 | — | — | — | 4532 | Per-token timing unavailable (not a request failure) |
| hf-002 | — | — | — | 5120 | Per-token timing unavailable (not a request failure) |
| hf-004 | — | — | — | 5120 | Per-token timing unavailable (not a request failure) |
| hf-008 | — | — | — | 5109 | Per-token timing unavailable (not a request failure) |
| hf-009 | — | — | — | 5120 | Per-token timing unavailable (not a request failure) |
| hf-013 | — | — | — | 4797 | Per-token timing unavailable (not a request failure) |
| hf-010 | — | — | — | 4500 | Per-token timing unavailable (not a request failure) |
| 5120 |
| 60.36ms |
| 1892 gaps for 5120 tokens |
| hf-003 | 148.89ms | 25.649s | 5120 | 37.55ms | 1741 gaps for 5120 tokens |
| hf-004 | 148.22ms | 26.034s | 5120 | 42.17ms | 1767 gaps for 5120 tokens |
| hf-005 | 144.71ms | 23.632s | 4472 | 19.86ms | 1607 gaps for 4472 tokens |
| hf-006 | 93.59ms | 20.704s | 4563 | 18.04ms | 1491 gaps for 4563 tokens |
| hf-007 | 92.67ms | 25.102s | 5120 | 44.30ms | 1802 gaps for 5120 tokens |
| hf-008 | 99.91ms | 25.201s | 5120 | 45.38ms | 1707 gaps for 5120 tokens |
| hf-009 | 125.54ms | 25.657s | 5120 | 45.40ms | 1749 gaps for 5120 tokens |
| hf-010 | 94.28ms | 26.962s | 5120 | 41.56ms | 1855 gaps for 5120 tokens |
| hf-011 | 73.51ms | 24.030s | 5120 | 41.90ms | 1726 gaps for 5120 tokens |
| hf-012 | 69.52ms | 20.784s | 4684 | 41.19ms | 1493 gaps for 4684 tokens |
| hf-013 | 83.26ms | 25.606s | 5120 | 19.45ms | 1774 gaps for 5120 tokens |
| hf-014 | 64.96ms | 22.831s | 5120 | 44.41ms | 1646 gaps for 5120 tokens |
| hf-015 | 69.93ms | 24.262s | 5120 | 18.83ms | 1800 gaps for 5120 tokens |
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 | Baseline char-16 | Drop | Limit | Result |
|---|---|---|---|---|---|
| hf-000 | 0.9068 | 0.9784 | 0.0716 | 0.1000 | Passed |
| hf-001 | 0.8678 | 0.7597 | -0.1081 | 0.1000 | Passed |
| hf-002 | 0.9396 | 0.8944 | -0.0452 | 0.1000 | Passed |
| hf-003 | 0.8514 | 0.9633 | 0.1120 | 0.1000 | Failed distinct 16-gram ratio 0.851 is more than 0.10 below baseline 0.963 over 21234 chars |
| hf-004 | 0.9576 | 0.9479 | -0.0097 | 0.1000 | Passed |
| hf-005 | 0.9217 | 0.9048 | -0.0169 | 0.1000 | Passed |
| hf-006 | 0.9163 | 0.9607 | 0.0444 | 0.1000 | Passed |
| hf-007 | 0.8971 | 0.8991 | 0.0020 | 0.1000 | Passed |
| hf-008 | 0.8264 | 0.9360 | 0.1096 | 0.1000 | Failed distinct 16-gram ratio 0.826 is more than 0.10 below baseline 0.936 over 23495 chars |
| hf-009 | 0.8822 | 0.9555 | 0.0734 | 0.1000 | Passed |
| hf-010 | 0.9526 | 0.9513 | -0.0013 | 0.1000 | Passed |
| hf-011 | 0.8861 | 0.8237 | -0.0624 | 0.1000 | Passed |
| hf-012 | 0.8052 | 0.9790 | 0.1738 | 0.1000 | Failed distinct 16-gram ratio 0.805 is more than 0.10 below baseline 0.979 over 17821 chars |
| hf-013 | 0.8997 | 0.9074 | 0.0078 | 0.1000 | Passed |
| hf-014 | 0.8017 | 0.9737 | 0.1721 | 0.1000 | Failed distinct 16-gram ratio 0.802 is more than 0.10 below baseline 0.974 over 20923 chars |
| hf-015 | 0.8499 | 0.8589 | 0.0090 | 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-000 | 1 | 0.919431 | 1 | 0 |
| hf-006 | 1 | 0.88093 | 1 | 0 |
| hf-001 | 1 | 0.202313 | 1 | 0 |
| hf-007 | 1 | 0.31014 | 1 | 0 |
| hf-012 | 1 | 0.867259 | 1 | 0 |
| hf-011 | 1 | 0.103972 | 1 | 0 |
| hf-014 | 1 | 0.902302 | 1 | 0 |
| hf-015 | 1 | 0.249512 | 1 | 0 |
| hf-005 | 1 | 0.297692 | 1 | 0 |
| hf-004 | 1 | 0.561 | 1 | 0 |
| hf-003 | 1 | 0.322659 | 1 | 0 |
| hf-002 | 1 | 0.111747 | 1 | 0 |
| hf-008 | 1 | 0.380167 | 1 | 0 |
| hf-009 | 1 | 0.415164 | 1 | 0 |
| hf-010 | 1 | 0.302381 | 1 | 0 |
| hf-013 | 1 | 0.260907 | 1 | 0 |
| hf-000 | 2 | 0.919431 | 1 | 0 |
| hf-006 | 2 | 0.88093 | 1 | 0 |
| hf-007 | 2 | 0.31014 | 1 | 0 |
| hf-001 | 2 | 0.202313 | 1 | 0 |
| hf-012 | 2 | 0.867259 | 1 | 0 |
| hf-011 | 2 | 0.103972 | 1 | 0 |
| hf-015 | 2 | 0.249512 | 1 | 0 |
| hf-014 | 2 | 0.902302 | 1 | 0 |
| hf-005 | 2 | 0.297692 | 1 | 0 |
| hf-003 | 2 | 0.322659 | 1 | 0 |
| hf-004 | 2 | 0.561 | 1 | 0 |
| hf-002 | 2 | 0.111747 | 1 | 0 |
| hf-008 | 2 | 0.380167 | 1 | 0 |
| hf-009 | 2 | 0.415164 | 1 | 0 |
| hf-013 | 2 | 0.260907 | 1 | 0 |
| hf-010 | 2 | 0.302381 | 1 | 0 |
| hf-006 | 3 | 0.88093 | 1 | 0 |
| hf-007 | 3 | 0.31014 | 1 | 0 |
| hf-000 | 3 | 0.919431 | 1 | 0 |
| hf-001 | 3 | 0.202313 | 1 | 0 |
| hf-012 | 3 | 0.867259 | 1 | 0 |
| hf-011 | 3 | 0.103972 | 1 | 0 |
| hf-014 | 3 | 0.902302 | 1 | 0 |
| hf-015 | 3 | 0.249512 | 1 | 0 |
| hf-005 | 3 | 0.297692 | 1 | 0 |
| hf-004 | 3 | 0.561 | 1 | 0 |
| hf-003 | 3 | 0.322659 | 1 | 0 |
| hf-002 | 3 | 0.111747 | 1 | 0 |
| hf-008 | 3 | 0.380167 | 1 | 0 |
| hf-009 | 3 | 0.415164 | 1 | 0 |
| hf-010 | 3 | 0.302381 | 1 | 0 |
| hf-013 | 3 | 0.260907 | 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