Round score
0.1136
weighted tier speedup after failure cap and deduction
Eligible requests
13
trusted baseline exclusions removed
Failed eligible requests
10
each request counted once
Raw weighted speedup
+19.05%
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
Request failures cap speed credit at zero: min(0.190495, 0) = 0.190495. Negative speedups remain negative.
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.
13 prompts
| Prompt | Diagnostic speedup | Baseline | Candidate | Aligned output | Diagnostic / failure |
|---|---|---|---|---|---|
| hf-015 | — | — | — | 4564 | Per-token timing unavailable (not a request failure) |
| hf-002 | — | — | — | 3601 | candidate output below tolerance |
| hf-004 | — | — | — |
13 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 612.34ms | 66.616s | 5120 | 33.30ms | 2752 gaps for 5120 tokens |
| hf-001 | 619.48ms | 54.305s | 3759 | 37.48ms | 2102 gaps for 3759 tokens |
| hf-002 | 609.61ms | 43.458s |
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 | C16 | 13 | 13 | 9.43 |
| 2 | 0.000s | C16 | 13 | 13 | 10.08 |
| 3 | 0.000s | C16 | 13 | 13 | 9.04 |
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 |
|---|
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 |
|---|---|---|---|---|---|---|
| 2k | 3 | 25% | 73.870s | 59.642s | +19.26% | +4.82% |
| 4k | 3 | 25% | 78.008s | 63.263s | +18.90% | +4.73% |
| 8k | 3 | 25% | 71.897s | 60.462s | +15.90% | +3.98% |
| 16k | 4 | 25% | 85.556s | 66.621s | +22.13% | +5.53% |
| 3983 |
| candidate output below tolerance |
| hf-006 | — | — | — | 2825 | candidate output below tolerance |
| hf-007 | — | — | — | 2720 | candidate output below tolerance |
| hf-010 | — | — | — | 3189 | candidate output below tolerance |
| hf-013 | — | — | — | 3412 | candidate output below tolerance |
| hf-008 | — | — | — | 4442 | candidate output below tolerance |
| hf-003 | — | — | — | 4358 | candidate output below tolerance |
| hf-014 | — | — | — | 5102 | Per-token timing unavailable (not a request failure) |
| hf-000 | — | — | — | 4642 | Per-token timing unavailable (not a request failure) |
| hf-001 | — | — | — | 3759 | candidate output below tolerance |
| hf-009 | — | — | — | 3635 | candidate output below tolerance |
| 3601 |
| 33.17ms |
| 1562 gaps for 3601 tokens |
| hf-003 | 606.14ms | 59.642s | 4358 | 33.25ms | 2235 gaps for 4358 tokens |
| hf-004 | 603.67ms | 48.517s | 3983 | 36.35ms | 1816 gaps for 3983 tokens |
| hf-006 | 613.40ms | 35.221s | 2825 | 33.85ms | 1215 gaps for 2825 tokens |
| hf-007 | 617.41ms | 32.225s | 2720 | 36.81ms | 1123 gaps for 2720 tokens |
| hf-008 | 614.96ms | 60.945s | 4442 | 33.59ms | 2314 gaps for 4442 tokens |
| hf-009 | 603.75ms | 56.949s | 3635 | 33.30ms | 2127 gaps for 3635 tokens |
| hf-010 | 604.54ms | 38.000s | 3189 | 32.91ms | 1347 gaps for 3189 tokens |
| hf-013 | 608.82ms | 45.003s | 3412 | 37.12ms | 1657 gaps for 3412 tokens |
| hf-014 | 604.51ms | 60.445s | 5120 | 33.26ms | 2275 gaps for 5120 tokens |
| hf-015 | 607.64ms | 53.402s | 4564 | 35.26ms | 1954 gaps for 4564 tokens |
| Baseline char-16 |
|---|
| Drop |
|---|
| Limit |
|---|
| Result |
|---|
| hf-000 | 0.9482 | 0.9474 | -0.0009 | 0.1000 | Passed |
| hf-001 | 0.8487 | 0.9421 | 0.0934 | 0.1000 | Passed |
| hf-002 | 0.9815 | 0.8927 | -0.0889 | 0.1000 | Passed |
| hf-003 | 0.9738 | 0.9506 | -0.0233 | 0.1000 | Passed |
| hf-004 | 0.9720 | 0.6318 | -0.3402 | 0.1000 | Passed |
| hf-006 | 0.9343 | 0.7474 | -0.1869 | 0.1000 | Passed |
| hf-007 | 0.9533 | 0.9622 | 0.0089 | 0.1000 | Passed |
| hf-008 | 0.9531 | 0.8891 | -0.0640 | 0.1000 | Passed |
| hf-009 | 0.9621 | 0.9256 | -0.0364 | 0.1000 | Passed |
| hf-010 | 0.9691 | 0.8965 | -0.0727 | 0.1000 | Passed |
| hf-013 | 0.9730 | 0.9311 | -0.0419 | 0.1000 | Passed |
| hf-014 | 0.9691 | 0.6795 | -0.2896 | 0.1000 | Passed |
| hf-015 | 0.8975 | 0.9589 | 0.0614 | 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-015 | 1 | 0.683513 | 1 | 0 |
| hf-007 | 1 | 0.984061 | 1 | 0 |
| hf-006 | 1 | 0.943683 | 1 | 0 |
| hf-010 | 1 | 0.868359 | 1 | 0 |
| hf-002 | 1 | 0.927754 | 1 | 0 |
| hf-004 | 1 | 0.800229 | 1 | 0 |
| hf-009 | 1 | 0.210266 | 1 | 0 |
| hf-008 | 1 | 0.428242 | 1 | 0 |
| hf-014 | 1 | 0.704503 | 1 | 0 |
| hf-013 | 1 | 0.684286 | 1 | 0 |
| hf-003 | 1 | 0.581732 | 1 | 0 |
| hf-000 | 1 | 0.285333 | 1 | 0 |
| hf-001 | 1 | 0.241279 | 1 | 0 |
| hf-013 | 2 | 0.684286 | 1 | 0 |
| hf-010 | 2 | 0.868359 | 1 | 0 |
| hf-006 | 2 | 0.943683 | 1 | 0 |
| hf-001 | 2 | 0.241279 | 1 | 0 |
| hf-004 | 2 | 0.800229 | 1 | 0 |
| hf-015 | 2 | 0.683513 | 1 | 0 |
| hf-003 | 2 | 0.581732 | 1 | 0 |
| hf-007 | 2 | 0.984061 | 1 | 0 |
| hf-014 | 2 | 0.704503 | 1 | 0 |
| hf-009 | 2 | 0.210266 | 1 | 0 |
| hf-002 | 2 | 0.927754 | 1 | 0 |
| hf-008 | 2 | 0.428242 | 1 | 0 |
| hf-000 | 2 | 0.285333 | 1 | 0 |
| hf-002 | 3 | 0.927754 | 1 | 0 |
| hf-003 | 3 | 0.581732 | 1 | 0 |
| hf-006 | 3 | 0.943683 | 1 | 0 |
| hf-007 | 3 | 0.984061 | 1 | 0 |
| hf-014 | 3 | 0.704503 | 1 | 0 |
| hf-013 | 3 | 0.684286 | 1 | 0 |
| hf-004 | 3 | 0.800229 | 1 | 0 |
| hf-009 | 3 | 0.210266 | 1 | 0 |
| hf-001 | 3 | 0.241279 | 1 | 0 |
| hf-010 | 3 | 0.868359 | 1 | 0 |
| hf-008 | 3 | 0.428242 | 1 | 0 |
| hf-015 | 3 | 0.683513 | 1 | 0 |
| hf-000 | 3 | 0.285333 | 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