Round score
0
weighted tier speedup after failure cap and deduction
Eligible requests
—
trusted baseline exclusions removed
Failed eligible requests
—
each request counted once
Raw weighted speedup
—
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.
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.
Reference engine
The baseline is what candidates are measured against, so it has no speedup of its own. Its SLA replay is below.
28 requests
| Request | TTFT | Total | Tokens | Slowest gap | Stream |
|---|---|---|---|---|---|
| hf-000 | 743.97ms | 105.343s | 5120 | 41.56ms | 3053 gaps for 5120 tokens |
| hf-001 | 755.84ms | 110.025s | 5120 | 64.01ms | 3319 gaps for 5120 tokens |
| hf-002 | 742.24ms |
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 | 32 | 32 | 24.81 |
| 2 | 0.000s | C32 | 32 | 32 | 24.51 |
| 3 | 0.000s | C32 | 32 | 32 | 24.85 |
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 |
|---|
| 92.182s |
| 4583 |
| 41.56ms |
| 2439 gaps for 4583 tokens |
| hf-003 | 736.25ms | 94.222s | 5120 | 41.54ms | 2510 gaps for 5120 tokens |
| hf-004 | 721.61ms | 104.271s | 4799 | 42.72ms | 2863 gaps for 4799 tokens |
| hf-005 | 753.80ms | 93.950s | 5120 | 64.25ms | 2494 gaps for 5120 tokens |
| hf-006 | 728.23ms | 81.421s | 5120 | 42.01ms | 2133 gaps for 5120 tokens |
| hf-007 | 720.63ms | 92.579s | 5120 | 41.95ms | 2470 gaps for 5120 tokens |
| hf-008 | 661.04ms | 70.753s | 3585 | 110.66ms | 1814 gaps for 3585 tokens |
| hf-009 | 749.89ms | 100.821s | 4708 | 64.22ms | 2755 gaps for 4708 tokens |
| hf-010 | 741.46ms | 103.029s | 5120 | 64.11ms | 2858 gaps for 5120 tokens |
| hf-013 | 747.64ms | 89.560s | 5120 | 64.10ms | 2357 gaps for 5120 tokens |
| hf-014 | 746.18ms | 104.079s | 5120 | 41.43ms | 2975 gaps for 5120 tokens |
| hf-015 | 745.04ms | 68.001s | 3007 | 41.39ms | 1746 gaps for 3007 tokens |
| hf-018 | 729.18ms | 92.287s | 5120 | 42.51ms | 2443 gaps for 5120 tokens |
| hf-019 | 737.56ms | 104.540s | 5120 | 64.52ms | 2937 gaps for 5120 tokens |
| hf-020 | 739.64ms | 108.692s | 5120 | 64.51ms | 3204 gaps for 5120 tokens |
| hf-021 | 747.84ms | 71.469s | 3083 | 64.10ms | 1834 gaps for 3083 tokens |
| hf-022 | 743.63ms | 97.791s | 5120 | 64.41ms | 2625 gaps for 5120 tokens |
| hf-023 | 734.31ms | 101.844s | 5120 | 42.52ms | 2853 gaps for 5120 tokens |
| hf-024 | 738.55ms | 97.107s | 4546 | 64.39ms | 2600 gaps for 4546 tokens |
| hf-025 | 707.83ms | 89.618s | 5120 | 42.87ms | 2375 gaps for 5120 tokens |
| hf-026 | 738.36ms | 89.984s | 4031 | 64.62ms | 2370 gaps for 4031 tokens |
| hf-027 | 734.62ms | 71.930s | 3503 | 42.98ms | 1853 gaps for 3503 tokens |
| hf-028 | 709.70ms | 92.291s | 5120 | 42.31ms | 2444 gaps for 5120 tokens |
| hf-029 | 733.06ms | 98.504s | 4943 | 64.64ms | 2653 gaps for 4943 tokens |
| hf-030 | 733.05ms | 67.456s | 4087 | 43.44ms | 1732 gaps for 4087 tokens |
| hf-031 | 710.47ms | 84.541s | 4162 | 42.94ms | 2209 gaps for 4162 tokens |
| Top p |
|---|
| Seed |
|---|
| hf-018 | 1 | 0.746916 | 1 | 0 |
| hf-015 | 1 | 0.861071 | 1 | 0 |
| hf-027 | 1 | 0.343957 | 1 | 0 |
| hf-006 | 1 | 0.79037 | 1 | 0 |
| hf-011 | 1 | 0.183035 | 1 | 0 |
| hf-030 | 1 | 0.836317 | 1 | 0 |
| hf-017 | 1 | 0.44517 | 1 | 0 |
| hf-025 | 1 | 0.941279 | 1 | 0 |
| hf-008 | 1 | 0.670483 | 1 | 0 |
| hf-021 | 1 | 0.174302 | 1 | 0 |
| hf-031 | 1 | 0.494802 | 1 | 0 |
| hf-012 | 1 | 0.864593 | 1 | 0 |
| hf-013 | 1 | 0.84917 | 1 | 0 |
| hf-026 | 1 | 0.289506 | 1 | 0 |
| hf-005 | 1 | 0.599106 | 1 | 0 |
| hf-016 | 1 | 0.624319 | 1 | 0 |
| hf-003 | 1 | 0.602455 | 1 | 0 |
| hf-028 | 1 | 0.696372 | 1 | 0 |
| hf-023 | 1 | 0.408635 | 1 | 0 |
| hf-007 | 1 | 0.735279 | 1 | 0 |
| hf-024 | 1 | 0.274026 | 1 | 0 |
| hf-002 | 1 | 0.993342 | 1 | 0 |
| hf-022 | 1 | 0.477162 | 1 | 0 |
| hf-029 | 1 | 0.257657 | 1 | 0 |
| hf-009 | 1 | 0.355006 | 1 | 0 |
| hf-010 | 1 | 0.347568 | 1 | 0 |
| hf-014 | 1 | 0.294577 | 1 | 0 |
| hf-000 | 1 | 0.312286 | 1 | 0 |
| hf-019 | 1 | 0.279221 | 1 | 0 |
| hf-004 | 1 | 0.148535 | 1 | 0 |
| hf-020 | 1 | 0.156961 | 1 | 0 |
| hf-001 | 1 | 0.122854 | 1 | 0 |
| hf-013 | 2 | 0.84917 | 1 | 0 |
| hf-016 | 2 | 0.624319 | 1 | 0 |
| hf-021 | 2 | 0.174302 | 1 | 0 |
| hf-008 | 2 | 0.670483 | 1 | 0 |
| hf-028 | 2 | 0.696372 | 1 | 0 |
| hf-017 | 2 | 0.44517 | 1 | 0 |
| hf-024 | 2 | 0.274026 | 1 | 0 |
| hf-012 | 2 | 0.864593 | 1 | 0 |
| hf-002 | 2 | 0.993342 | 1 | 0 |
| hf-006 | 2 | 0.79037 | 1 | 0 |
| hf-015 | 2 | 0.861071 | 1 | 0 |
| hf-030 | 2 | 0.836317 | 1 | 0 |
| hf-003 | 2 | 0.602455 | 1 | 0 |
| hf-026 | 2 | 0.289506 | 1 | 0 |
| hf-025 | 2 | 0.941279 | 1 | 0 |
| hf-031 | 2 | 0.494802 | 1 | 0 |
| hf-007 | 2 | 0.735279 | 1 | 0 |
| hf-005 | 2 | 0.599106 | 1 | 0 |
| hf-022 | 2 | 0.477162 | 1 | 0 |
| hf-027 | 2 | 0.343957 | 1 | 0 |
| hf-004 | 2 | 0.148535 | 1 | 0 |
| hf-023 | 2 | 0.408635 | 1 | 0 |
| hf-010 | 2 | 0.347568 | 1 | 0 |
| hf-014 | 2 | 0.294577 | 1 | 0 |
| hf-000 | 2 | 0.312286 | 1 | 0 |
| hf-018 | 2 | 0.746916 | 1 | 0 |
| hf-029 | 2 | 0.257657 | 1 | 0 |
| hf-009 | 2 | 0.355006 | 1 | 0 |
| hf-019 | 2 | 0.279221 | 1 | 0 |
| hf-011 | 2 | 0.183035 | 1 | 0 |
| hf-001 | 2 | 0.122854 | 1 | 0 |
| hf-020 | 2 | 0.156961 | 1 | 0 |
| hf-017 | 3 | 0.44517 | 1 | 0 |
| hf-011 | 3 | 0.183035 | 1 | 0 |
| hf-010 | 3 | 0.347568 | 1 | 0 |
| hf-007 | 3 | 0.735279 | 1 | 0 |
| hf-030 | 3 | 0.836317 | 1 | 0 |
| hf-015 | 3 | 0.861071 | 1 | 0 |
| hf-027 | 3 | 0.343957 | 1 | 0 |
| hf-019 | 3 | 0.279221 | 1 | 0 |
| hf-012 | 3 | 0.864593 | 1 | 0 |
| hf-029 | 3 | 0.257657 | 1 | 0 |
| hf-008 | 3 | 0.670483 | 1 | 0 |
| hf-031 | 3 | 0.494802 | 1 | 0 |
| hf-005 | 3 | 0.599106 | 1 | 0 |
| hf-013 | 3 | 0.84917 | 1 | 0 |
| hf-016 | 3 | 0.624319 | 1 | 0 |
| hf-002 | 3 | 0.993342 | 1 | 0 |
| hf-018 | 3 | 0.746916 | 1 | 0 |
| hf-028 | 3 | 0.696372 | 1 | 0 |
| hf-025 | 3 | 0.941279 | 1 | 0 |
| hf-006 | 3 | 0.79037 | 1 | 0 |
| hf-004 | 3 | 0.148535 | 1 | 0 |
| hf-003 | 3 | 0.602455 | 1 | 0 |
| hf-022 | 3 | 0.477162 | 1 | 0 |
| hf-009 | 3 | 0.355006 | 1 | 0 |
| hf-020 | 3 | 0.156961 | 1 | 0 |
| hf-023 | 3 | 0.408635 | 1 | 0 |
| hf-024 | 3 | 0.274026 | 1 | 0 |
| hf-026 | 3 | 0.289506 | 1 | 0 |
| hf-014 | 3 | 0.294577 | 1 | 0 |
| hf-000 | 3 | 0.312286 | 1 | 0 |
| hf-021 | 3 | 0.174302 | 1 | 0 |
| hf-001 | 3 | 0.122854 | 1 | 0 |
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