Durchsatz
| Zeile | R | tok/s | TTFT s | Prompt max | Cache | Reasoning | Tool-Anteil | Aufrufe | gleichzeitig |
|---|---|---|---|---|---|---|---|---|---|
| V3R7-20260912T142811 | 7 | 53.7 | 2.79 | 91286 | 0% | 81% | 44% | 134 | 2 |
| v3-hermes-gpt-5.6-terra | 1 | 196525.3 | 4.42 | 26205 | 91% | 92% | 0% | 12 | 1 |
| v3-hermes-gpt-5.6-terra | 2 | 276230.9 | 6.11 | 42832 | 96% | 93% | 0% | 15 | 2 |
| v3-hermes-gpt-5.6-terra | 3 | 462078.8 | 7.25 | 66964 | 96% | 92% | 0% | 14 | 1 |
| v3-hermes-gpt-5.6-terra | 4 | 456607.8 | 8.78 | 79537 | 97% | 86% | 0% | 9 | 1 |
| v3-hermes-gpt-5.6-terra | 5 | 368113.6 | 6.07 | 87069 | 98% | 80% | 0% | 6 | 1 |
| v3-hermes-gpt-5.6-terra | 6 | 360322.3 | 6.31 | 96556 | 98% | 85% | 0% | 8 | 1 |
| v3-hermes-gpt-5.6-terra | 7 | 348478.9 | 6.98 | 101889 | 98% | 70% | 0% | 5 | 1 |
| v3-hermes-gpt-5.6-terra | 1 | – | – | 0 | – | – | – | 0 | 0 |
| v3-hermes-gpt-5.6-terra | 2 | – | – | 0 | – | – | – | 0 | 0 |
| v3-hermes-gpt-5.6-sol | 1 | 299732.2 | 136.29 | 15379 | 0% | 100% | 0% | 3 | 2 |
| v3-hermes-gpt-5.6-sol | 2 | 425511.0 | 16.04 | 23409 | 94% | 100% | 0% | 6 | 2 |
| v3-hermes-gpt-5.6-sol | 3 | – | 210.69 | 0 | – | – | 0% | 2 | 2 |
| v3-hermes-gpt-5.6-sol | 4 | – | 232.49 | 0 | – | – | 0% | 2 | 2 |
| v3-hermes-gpt-5.6-sol | 5 | – | 201.82 | 0 | – | – | 0% | 2 | 2 |
| v3-hermes-gpt-5.6-sol | 6 | – | 282.19 | 0 | – | – | 0% | 1 | 2 |
| v3-hermes-gpt-5.6-sol | 7 | – | 285.38 | 0 | – | – | 0% | 1 | 3 |
| v3-hermes-gpt-5.6-sol | 1 | – | – | 0 | – | – | – | 0 | 0 |
| v3-hermes-gpt-5.6-luna | 1 | 338476.2 | 5.53 | 30819 | 93% | 92% | 0% | 14 | 1 |
| v3-hermes-gpt-5.6-luna | 2 | 416716.2 | 6.52 | 70144 | 92% | 94% | 0% | 19 | 2 |
| v3-hermes-gpt-5.6-luna | 3 | 419596.4 | 5.96 | 115032 | 97% | 96% | 0% | 24 | 1 |
| v3-hermes-gpt-5.6-luna | 4 | 334476.5 | 6.85 | 176239 | 98% | 97% | 1% | 31 | 1 |
| v3-hermes-gpt-5.6-luna | 5 | 331708.4 | 6.36 | 211756 | 98% | 93% | 1% | 15 | 1 |
| v3-hermes-gpt-5.6-luna | 6 | 426879.3 | 8.65 | 232687 | 99% | 90% | 1% | 11 | 1 |
| v3-hermes-gpt-5.6-luna | 7 | 473009.8 | 8.01 | 255312 | 99% | 86% | 1% | 8 | 1 |
| v3-hermes-gpt-5.6-luna | 1 | – | – | 0 | – | – | – | 0 | 0 |
| v3-hermes-gpt-5.6-luna | 2 | – | – | 0 | – | – | – | 0 | 0 |
| v3-hermes-qwen-rec3-131k | 1 | 59.3 | 2.44 | 89981 | 0% | 80% | 38% | 137 | 2 |
| v3-hermes-qwen-rec3-131k | 2 | 54.5 | 2.54 | 89612 | 0% | 81% | 48% | 124 | 3 |
| v3-hermes-qwen-rec3-131k | 3 | 58.1 | 3.02 | 89766 | 0% | 80% | 49% | 104 | 2 |
| v3-hermes-qwen-rec3-131k | 4 | 55.5 | 2.83 | 89110 | 0% | 87% | 40% | 157 | 2 |
| v3-hermes-qwen-rec3-131k | 5 | 58.9 | 3.13 | 89971 | 0% | 80% | 46% | 155 | 2 |
| v3-hermes-qwen-rec3-131k | 6 | 54.5 | 3.09 | 90059 | 0% | 84% | 42% | 224 | 3 |
| v3-hermes-qwen-rec3-131k | 7 | 55.5 | 2.35 | 92270 | 0% | 83% | 41% | 317 | 3 |
| v3-hermes-qwen-rec3-131k | 1 | 67.4 | 2.25 | 70692 | 0% | 88% | 38% | 49 | 1 |
| v3-hermes-qwen-rec3-131k | 2 | 56.4 | 2.39 | 89736 | 0% | 84% | 41% | 146 | 2 |
| v3-hermes-qwen-rec3-131k | 3 | 56.1 | 2.62 | 89574 | 0% | 82% | 45% | 141 | 2 |
| v3-hermes-qwen-rec3-131k | 4 | 47.7 | 2.88 | 128459 | 0% | 86% | 53% | 241 | 2 |
| v3-hermes-qwen-rec3 | 1 | 112.4 | 2.13 | 49928 | 0% | 79% | 26% | 155 | 1 |
| v3-hermes-qwen-rec3 | 2 | 108.8 | 3.21 | 57629 | 0% | 83% | 32% | 68 | 1 |
| v3-hermes-qwen-rec3 | 3 | 117.9 | 2.19 | 50275 | 0% | 78% | 28% | 355 | 1 |
| v3-hermes-qwen-rec3 | 4 | 110.2 | 5.15 | 60931 | 0% | 80% | 25% | 35 | 1 |
| v3-hermes-qwen-stock | 1 | 116.5 | 2.34 | 56851 | 0% | 80% | 26% | 108 | 1 |
| v3-hermes-qwen-stock | 2 | 115.7 | 2.47 | 64333 | 0% | 80% | 32% | 82 | 1 |
| v3-hermes-qwen-stock | 3 | 119.3 | 2.15 | 55949 | 0% | 80% | 30% | 327 | 1 |
| v3-hermes-qwen-stock | 4 | 109.8 | 6.80 | 63302 | 0% | 80% | 45% | 53 | 1 |
| v3-hermes-qwen-stock | 5 | 122.2 | 4.51 | 56987 | 0% | 75% | 39% | 363 | 1 |
| v3-hermes-claude-sonnet-5 | 1 | 125.8 | 1.84 | 47347 | 92% | 65% | 8% | 26 | 1 |
| v3-hermes-claude-sonnet-5 | 2 | 116.3 | 1.76 | 111791 | 91% | 51% | 29% | 32 | 1 |
| v3-hermes-claude-sonnet-5 | 4 | 136.9 | 1.09 | 296604 | 98% | 25% | 55% | 51 | 1 |
| v3-hermes-claude-sonnet-5 | 5 | 143.8 | 1.23 | 363801 | 100% | 31% | 57% | 64 | 1 |
| v3-hermes-claude-sonnet-5 | 6 | 149.8 | 1.25 | 447067 | 100% | 29% | 58% | 66 | 1 |
| v3-hermes-claude-sonnet-5 | 7 | 142.9 | 1.13 | 499929 | 99% | 37% | 50% | 137 | 1 |
| v3-hermes-claude-opus-5 | 1 | 105.5 | 1.48 | 108039 | 90% | 84% | 15% | 52 | 1 |
| v3-hermes-claude-opus-5 | 3 | 106.0 | 1.95 | 336602 | 94% | 31% | 40% | 78 | 1 |
| v3-hermes-claude-opus-5 | 4 | 103.6 | 1.98 | 459767 | 98% | 26% | 43% | 57 | 1 |
| v3-hermes-claude-opus-5 | 5 | 82.8 | 1.14 | 239064 | 83% | 48% | 44% | 6 | 1 |
| v3-hermes-claude-opus-5 | 6 | 104.2 | 1.16 | 396128 | 96% | 2% | 47% | 92 | 1 |
| v3-hermes-claude-opus-5 | 7 | 102.1 | 1.11 | 498956 | 99% | 5% | 53% | 214 | 1 |
| v3-hermes-claude-haiku-4-5 | 1 | 169.9 | 0.83 | 38512 | 96% | 0% | 15% | 34 | 1 |
| v3-hermes-claude-haiku-4-5 | 2 | 165.4 | 0.83 | 74596 | 95% | 0% | 20% | 27 | 1 |
| v3-hermes-claude-haiku-4-5 | 3 | 108.7 | 0.83 | 106071 | 93% | 0% | 20% | 15 | 1 |
| v3-hermes-claude-haiku-4-5 | 5 | 122.0 | 0.82 | 140897 | 92% | 0% | 17% | 50 | 1 |
| v3-hermes-claude-haiku-4-5 | 6 | 147.1 | 0.88 | 83385 | 98% | 0% | 24% | 19 | 1 |
| v3-hermes-claude-haiku-4-5 | 7 | 112.3 | 0.90 | 111186 | 98% | 0% | 17% | 11 | 1 |
| v3-hermes-claude-fable-5-1 | 1 | 119.5 | 1.44 | 71183 | 94% | 49% | 19% | 26 | 1 |
| v3-hermes-claude-fable-5-1 | 2 | 118.8 | 1.91 | 121624 | 95% | 27% | 27% | 24 | 1 |
| v3-hermes-claude-fable-5-1 | 4 | 124.7 | 2.48 | 229748 | 97% | 52% | 38% | 41 | 1 |
| v3-hermes-claude-fable-5-1 | 5 | 145.9 | 3.03 | 277906 | 97% | 56% | 38% | 37 | 1 |
| v3-hermes-claude-fable-5-1 | 6 | 134.9 | 2.34 | 320413 | 88% | 41% | 44% | 26 | 1 |
| v3-hermes-claude-fable-5-1 | 7 | 127.2 | 2.55 | 349270 | 96% | 28% | 46% | 27 | 1 |
| v3-cc-sonnet-5 | 1 | 122.7 | 0.94 | 89158 | 96% | 0% | 5% | 33 | 2 |
| v3-cc-sonnet-5 | 2 | 123.3 | 0.93 | 168860 | 98% | 0% | 16% | 70 | 1 |
| v3-cc-sonnet-5 | 4 | 134.3 | 1.01 | 158013 | 98% | 0% | 45% | 54 | 1 |
| v3-cc-sonnet-5 | 5 | 127.6 | 0.96 | 165538 | 98% | 0% | 30% | 113 | 1 |
| v3-cc-sonnet-5 | 6 | 119.5 | 0.98 | 167527 | 98% | 0% | 35% | 150 | 1 |
| v3-cc-sonnet-5 | 7 | 122.8 | 1.00 | 148921 | 94% | 0% | 42% | 18 | 1 |
| v3-cc-opus-5 | 2 | 112.6 | 1.54 | 158702 | 97% | 0% | 34% | 87 | 1 |
| v3-cc-opus-5 | 3 | 101.2 | 1.12 | 167127 | 99% | 0% | 55% | 190 | 1 |
| v3-cc-opus-5 | 4 | 102.4 | 1.24 | 166999 | 98% | 0% | 82% | 79 | 1 |
| v3-cc-opus-5 | 5 | 97.9 | 1.20 | 168535 | 98% | 0% | 74% | 114 | 1 |
| v3-cc-opus-5 | 6 | 84.5 | 0.92 | 94642 | 97% | 0% | 29% | 35 | 1 |
| v3-cc-opus-5 | 7 | 143.3 | 0.91 | 168047 | 100% | 0% | 36% | 446 | 1 |
| v3-cc-haiku-4-5 | 1 | 102.7 | 1.38 | 51970 | 39% | 0% | 0% | 3 | 2 |
| v3-cc-haiku-4-5 | 2 | 189.6 | 1.38 | 109846 | 93% | 0% | 1% | 11 | 1 |
| v3-cc-haiku-4-5 | 4 | 180.5 | 1.65 | 152244 | 99% | 0% | 3% | 17 | 1 |
| v3-cc-haiku-4-5 | 5 | 193.1 | 1.69 | 171010 | 100% | 0% | 4% | 29 | 1 |
| v3-cc-haiku-4-5 | 6 | 195.7 | 1.69 | 187803 | 100% | 0% | 5% | 17 | 1 |
| v3-cc-haiku-4-5 | 7 | 186.6 | 1.40 | 196255 | 99% | 0% | 33% | 85 | 1 |
| v3-cc-fable-5-1 | 1 | 88.7 | 1.34 | 128030 | 94% | 0% | 34% | 24 | 2 |
| v3-cc-fable-5-1 | 3 | 83.4 | 2.25 | 205696 | 97% | 0% | 83% | 7 | 1 |
| v3-cc-fable-5-1 | 4 | 85.7 | 2.02 | 254007 | 98% | 0% | 85% | 11 | 1 |
| v3-cc-fable-5-1 | 5 | 86.0 | 2.13 | 280557 | 99% | 0% | 85% | 7 | 1 |
| v3-cc-fable-5-1 | 6 | 86.0 | 2.15 | 315935 | 99% | 0% | 87% | 9 | 1 |
| v3-cc-fable-5-1 | 7 | 81.8 | 2.43 | 391639 | 95% | 0% | 89% | 22 | 1 |
gelb: ≥ 4 gleichzeitig offene Streams beim Aufruf — tok/s und TTFT stehen dann unter Last.
Zeitstrahl
Titelbild-Wand
claude-code · claude-fable-5-1
claude-code · claude-haiku-4-5-20251001
claude-code · claude-opus-5
claude-code · claude-sonnet-5
hermes · claude-fable-5-1 · prune aus
hermes · claude-haiku-4-5-20251001 · prune aus
hermes · claude-opus-5 · prune aus
hermes · claude-sonnet-5 · prune aus
hermes · Qwen3.8-27B-Instruct · ctx 65.536 · prune aus
hermes · Qwen3.8-27B-Instruct · ctx 65.536 · prune aus
hermes · Qwen3.8-27B-Instruct · ctx 65.536 · prune aus
hermes · Qwen3.8-27B-Instruct · ctx 131.072 · prune aus
hermes · Qwen3.8-27B-Instruct · ctx 131.072 · prune aus
hermes · gpt-5.6-luna · prune aus
hermes · gpt-5.6-luna · prune aus
hermes · gpt-5.6-sol · prune aus
hermes · gpt-5.6-sol · prune aus
hermes · gpt-5.6-terra · prune aus
hermes · gpt-5.6-terra · prune aus
V3R7-20260912T142811Gesamt
20Läufe
104Runden
42.2Stunden Agentenzeit
186.9 MPrompt-Token
6.09 Merzeugte Token
V3R7-20260912T142811
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 7 | 5/5 | 16 16–17/24 | 57 | 134 | 8,078,039 | 136,771 | 1,020 | 0.931 | ok | 0cf29b88 |
     | ||||||||||
V3G56T-20260912T155032
hermes · gpt-5.6-terra · prune aus
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 5/5 | 9/24 | 3 | backend=remote: no local model, no vLLM counters | ok | 71def7fa | ||||
      | ||||||||||
| 2 | 5/5 | 11/24 | 8 | backend=remote: no local model, no vLLM counters | ok | 55ebcec4 | ||||
      | ||||||||||
| 3 | 5/5 | 11/24 | 3 | backend=remote: no local model, no vLLM counters | ok | b3be2f83 | ||||
     | ||||||||||
| 4 | 5/5 | 14 13–14/24 | 2 | backend=remote: no local model, no vLLM counters | ok | f1acc75f | ||||
      | ||||||||||
| 5 | 5/5 | 14 13–15/24 | 2 | backend=remote: no local model, no vLLM counters | ok | 4842a5d1 | ||||
      | ||||||||||
| 6 | 5/5 | 18 14–18/24 | 2 | backend=remote: no local model, no vLLM counters | ok | 8b5d09ee | ||||
      | ||||||||||
| 7 | 4/5 | 14 13–15/24 | 1 | backend=remote: no local model, no vLLM counters | ok | 5711c580 | ||||
     | ||||||||||
V3G56T-20260912T143007
hermes · gpt-5.6-terra · prune aus
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 5/5 | 9/24 | 3 | backend=remote: no local model, no vLLM counters | ok | 7133d6fe | ||||
      keine Proxy-Daten | ||||||||||
| 2 | 4/5 | 9 8–9/24 | 3 | backend=remote: no local model, no vLLM counters | ok | 25192dc9 | ||||
     keine Proxy-Daten | ||||||||||
V3G56S-20260912T155032
hermes · gpt-5.6-sol · prune aus
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 0/5 | —/24 | 6 | backend=remote: no local model, no vLLM counters | harness-failure | e3b0c442 | ||||
harness-failure | ||||||||||
| 2 | 1/5 | 0/24 | 7 | backend=remote: no local model, no vLLM counters | harness-failure | 834c9019 | ||||
   harness-failure | ||||||||||
| 3 | 1/5 | 0/24 | 6 | backend=remote: no local model, no vLLM counters | harness-failure | 834c9019 unverändert | ||||
   harness-failure | ||||||||||
| 4 | 1/5 | 0/24 | 6 | backend=remote: no local model, no vLLM counters | harness-failure | 834c9019 unverändert | ||||
   harness-failure | ||||||||||
| 5 | 1/5 | 0/24 | 6 | backend=remote: no local model, no vLLM counters | harness-failure | 834c9019 unverändert | ||||
   harness-failure | ||||||||||
| 6 | 1/5 | 0/24 | 6 | backend=remote: no local model, no vLLM counters | harness-failure | 834c9019 unverändert | ||||
   harness-failure | ||||||||||
| 7 | 1/5 | 0/24 | 6 | backend=remote: no local model, no vLLM counters | harness-failure | 834c9019 unverändert | ||||
  harness-failure | ||||||||||
V3G56S-20260912T143007
hermes · gpt-5.6-sol · prune aus
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 5/5 | 7/24 | 7 | backend=remote: no local model, no vLLM counters | ok | 364191b2 | ||||
     keine Proxy-Daten | ||||||||||
V3G56L-20260912T155032
hermes · gpt-5.6-luna · prune aus
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 4/5 | 7 7–8/24 | 3 | backend=remote: no local model, no vLLM counters | ok | de1f5ede | ||||
      | ||||||||||
| 2 | 5/5 | 9 8–9/24 | 7 | backend=remote: no local model, no vLLM counters | ok | 7f935395 | ||||
      | ||||||||||
| 3 | 5/5 | 9 8–9/24 | 5 | backend=remote: no local model, no vLLM counters | ok | eb98fcf0 | ||||
      | ||||||||||
| 4 | 5/5 | 10 9–10/24 | 5 | backend=remote: no local model, no vLLM counters | ok | 35596f05 | ||||
      | ||||||||||
| 5 | 5/5 | 10 9–10/24 | 3 | backend=remote: no local model, no vLLM counters | ok | ba035f12 | ||||
      | ||||||||||
| 6 | 4/5 | 10 10–11/24 | 2 | backend=remote: no local model, no vLLM counters | ok | c2d017f6 | ||||
      | ||||||||||
| 7 | 5/5 | 11 10–11/24 | 2 | backend=remote: no local model, no vLLM counters | ok | db60cb84 | ||||
     | ||||||||||
V3G56L-20260912T143007
hermes · gpt-5.6-luna · prune aus
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 5/5 | 9/24 | 5 | backend=remote: no local model, no vLLM counters | ok | ad47735b | ||||
      keine Proxy-Daten | ||||||||||
| 2 | 5/5 | 9/24 | 5 | backend=remote: no local model, no vLLM counters | ok | 8cf3a8ab | ||||
     keine Proxy-Daten | ||||||||||
V3F-20260911T213540
hermes · Qwen3.8-27B-Instruct · ctx 131.072 · prune aus
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 5/5 | 9 9–10/24 | 86 | 135 | 8,048,229 | 228,559 | 1,693 | 0.955 | ok | 6e84f1f1 |
      400400 | ||||||||||
| 2 | 5/5 | 10/24 | 67 | 125 | 8,182,219 | 164,785 | 1,318 | 0.954 | ok | bc0350f5 |
      | ||||||||||
| 3 | 5/5 | 11 10–11/24 | 48 | 105 | 6,621,229 | 131,926 | 1,256 | 0.937 | ok | e6713fcf |
      | ||||||||||
| 4 | 5/5 | 14 13–15/24 | 95 | 157 | 9,580,251 | 255,590 | 1,627 | 0.94 | ok | 2bc453ad |
      | ||||||||||
| 5 | 5/5 | 13 12–13/24 | 61 | 156 | 9,693,790 | 170,192 | 1,090 | 0.939 | ok | 02e6b8d0 |
      | ||||||||||
| 6 | 5/5 | 14 14–17/24 | 118 | 225 | 13,764,181 | 323,203 | 1,436 | 0.935 | ok | 8297b588 |
      | ||||||||||
| 7 | 5/5 | 17 15–17/24 | 180 | 318 | 19,154,329 | 327,228 | 1,029 | 0.946 | timeout/bad-dir | 428c695b |
     timeout/bad-dir | ||||||||||
V3-20260911T170309
hermes · Qwen3.8-27B-Instruct · ctx 131.072 · prune aus
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 5/5 | 7/24 | 32 | 49 | 2,412,533 | 108,562 | 2,215 | 0.943 | ok | 8e170e54 |
      | ||||||||||
| 2 | 5/5 | 10/24 | 63 | 147 | 8,435,023 | 158,021 | 1,074 | 0.953 | ok | bfc9ca7e |
      | ||||||||||
| 3 | 5/5 | 10/24 | 55 | 142 | 9,322,988 | 134,198 | 945 | 0.955 | ok | 44396803 |
      | ||||||||||
| 4 | 5/5 | 12/24 | 107 | 241 | 19,103,498 | 233,471 | 968 | 0.962 | harness-failure | c0314159 |
     400harness-failure | ||||||||||
V3-20260911T094854
hermes · Qwen3.8-27B-Instruct · ctx 65.536 · prune aus
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 5/5 | 12/24 | 58 | 155 | 6,160,941 | 322,568 | 2,081 | 0.83 | ok | 61efcd44 |
      | ||||||||||
| 2 | 5/5 | 12/24 | 35 | 68 | 2,918,247 | 145,410 | 2,138 | 0.725 | harness-failure | f894b7f1 |
      400harness-failure | ||||||||||
| 3 | 5/5 | 20/24 | 168 | 355 | 13,430,035 | 952,704 | 2,683 | 0.805 | ok | 4be76bb4 |
      | ||||||||||
| 4 | 5/5 | 20 19–21/24 | 28 | 35 | 1,448,044 | 141,297 | 4,037 | 0.607 | harness-failure | 4be76bb4 unverändert |
     400harness-failure | ||||||||||
V3-20260911T004703
hermes · Qwen3.8-27B-Instruct · ctx 65.536 · prune aus
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 5/5 | 8/24 | 67 | 108 | 4,655,783 | 251,599 | 2,329 | 0.852 | ok | 008831ad |
      | ||||||||||
| 2 | 5/5 | 8/24 | 41 | 81 | 3,486,562 | 158,477 | 1,956 | 0.844 | harness-failure | 49da2bb4 |
      400harness-failure | ||||||||||
| 3 | 5/5 | 9 8–9/24 | 180 | 327 | 14,892,206 | 590,534 | 1,805 | 0.869 | timeout/bad-dir | 60fe4b88 |
      timeout/bad-dir | ||||||||||
| 4 | 5/5 | 9 3–10/24 | 35 | 52 | 2,274,819 | 141,467 | 2,720 | 0.676 | harness-failure | 60fe4b88 unverändert |
     400harness-failure | ||||||||||
| 5 | 5/5 | 10 9–10/24 | 180 | 363 | 15,265,587 | 1,015,326 | 2,797 | 0.753 | timeout/bad-dir | b42066b0 |
     timeout/bad-dir | ||||||||||
V3-20260910T073640
hermes · Qwen3.8-27B-Instruct · ctx 65.536 · prune aus
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 1/5 | 3/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | 549dd73e |
  not observed (resumed) | ||||||||||
V3-20260910T073546--v3-hermes-claude-sonnet-5
hermes · claude-sonnet-5 · prune aus
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 3/5 | 7/24 | 6 | backend=remote: no local model, no vLLM counters | ok | 78d86201 | ||||
      | ||||||||||
| 2 | 5/5 | 12/24 | 15 | backend=remote: no local model, no vLLM counters | ok | 1d81d753 | ||||
      | ||||||||||
| 3 | 5/5 | 11 10–11/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | dbaa94ac |
      not observed (resumed) | ||||||||||
| 4 | 5/5 | 14/24 | 14 | backend=remote: no local model, no vLLM counters | ok | a51fb3a4 | ||||
      | ||||||||||
| 5 | 5/5 | 15 14–15/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | c6b1871b |
      not observed (resumed) | ||||||||||
| 6 | 5/5 | 20/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | 45181283 |
      not observed (resumed) | ||||||||||
| 7 | 5/5 | 18 13–19/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | 8c3bd906 |
     not observed (resumed) | ||||||||||
V3-20260910T073546--v3-hermes-claude-opus-5
hermes · claude-opus-5 · prune aus
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 5/5 | 9/24 | 29 | backend=remote: no local model, no vLLM counters | ok | 5e7273fe | ||||
      | ||||||||||
| 2 | 5/5 | 10 10–11/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | d8b4e43b |
      not observed (resumed) | ||||||||||
| 3 | 5/5 | 10/24 | 60 | backend=remote: no local model, no vLLM counters | ok | f54d5495 | ||||
      | ||||||||||
| 4 | 5/5 | 13/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | a309404f |
      not observed (resumed) | ||||||||||
| 5 | 5/5 | 14 13–15/24 | 6 | backend=remote: no local model, no vLLM counters | ok | 9b28b31e | ||||
      | ||||||||||
| 6 | 5/5 | 18 17–18/24 | 43 | backend=remote: no local model, no vLLM counters | ok | 316273d6 | ||||
      | ||||||||||
| 7 | 5/5 | 16 16–17/24 | 138 | backend=remote: no local model, no vLLM counters | ok | 953431c8 | ||||
     | ||||||||||
V3-20260910T073546--v3-hermes-claude-haiku-4-5
hermes · claude-haiku-4-5-20251001 · prune aus
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 1/5 | 4/24 | 3 | backend=remote: no local model, no vLLM counters | ok | 002ac637 | ||||
   | ||||||||||
| 2 | 1/5 | 4/24 | 4 | backend=remote: no local model, no vLLM counters | ok | 44322dc5 | ||||
   | ||||||||||
| 3 | 1/5 | 4/24 | 4 | backend=remote: no local model, no vLLM counters | ok | ddb08d58 | ||||
   | ||||||||||
| 4 | 1/5 | 4/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | 5930dd7c |
   not observed (resumed) | ||||||||||
| 5 | 1/5 | 4/24 | 6 | backend=remote: no local model, no vLLM counters | ok | 3b8db53e | ||||
   | ||||||||||
| 6 | 0/5 | 0/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | e6f349c4 |
   not observed (resumed) | ||||||||||
| 7 | 0/5 | 0/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | b00a0dda |
  not observed (resumed) | ||||||||||
V3-20260910T073546--v3-hermes-claude-fable-5-1
hermes · claude-fable-5-1 · prune aus
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 5/5 | 8/24 | 19 | backend=remote: no local model, no vLLM counters | ok | df2f808c | ||||
      | ||||||||||
| 2 | 5/5 | 9 8–9/24 | 14 | backend=remote: no local model, no vLLM counters | ok | 49b4b6ab | ||||
      | ||||||||||
| 3 | 5/5 | 9/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | 0b0d2bb4 |
      not observed (resumed) | ||||||||||
| 4 | 5/5 | 9 9–12/24 | 36 | backend=remote: no local model, no vLLM counters | ok | a5ad7f61 | ||||
    | ||||||||||
| 5 | 5/5 | 13/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | 34e80f00 |
      401not observed (resumed) | ||||||||||
| 6 | 5/5 | 17 13–17/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | a3886a02 |
      not observed (resumed) | ||||||||||
| 7 | 5/5 | 17 12–18/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | 9268a354 |
     not observed (resumed) | ||||||||||
V3-20260910T073546--v3-cc-sonnet-5
claude-code · claude-sonnet-5
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 5/5 | 7/24 | 9 | backend=remote: no local model, no vLLM counters | ok | 29b2661f | ||||
      | ||||||||||
| 2 | 5/5 | 10 9–10/24 | 24 | backend=remote: no local model, no vLLM counters | ok | 2b4dedb7 | ||||
      | ||||||||||
| 3 | 5/5 | 10/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | bb0d168c |
      not observed (resumed) | ||||||||||
| 4 | 5/5 | 10 10–11/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | bd8b6cae |
      not observed (resumed) | ||||||||||
| 5 | 5/5 | 12 11–12/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | a894a44f |
      not observed (resumed) | ||||||||||
| 6 | 5/5 | 11/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | 8baf0a6c |
      not observed (resumed) | ||||||||||
| 7 | 5/5 | 12/24 | 10 | backend=remote: no local model, no vLLM counters | ok | b59ab707 | ||||
     | ||||||||||
V3-20260910T073546--v3-cc-opus-5
claude-code · claude-opus-5
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 5/5 | 9/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | 34ddf81f |
      not observed (resumed) | ||||||||||
| 2 | 5/5 | 10/24 | 51 | backend=remote: no local model, no vLLM counters | ok | 407613f0 | ||||
      | ||||||||||
| 3 | 5/5 | 10/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | a4b1cbe2 |
      not observed (resumed) | ||||||||||
| 4 | 5/5 | 11 10–11/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | a6ef7889 |
      not observed (resumed) | ||||||||||
| 5 | 5/5 | 12/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | 2e3bf386 |
      not observed (resumed) | ||||||||||
| 6 | 5/5 | 12 12–13/24 | 38 | backend=remote: no local model, no vLLM counters | ok | 6198de05 | ||||
      | ||||||||||
| 7 | 5/5 | 12/24 | 86 | backend=remote: no local model, no vLLM counters | ok | d73958a0 | ||||
     | ||||||||||
V3-20260910T073546--v3-cc-haiku-4-5
claude-code · claude-haiku-4-5-20251001
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 2/5 | 6/24 | 2 | backend=remote: no local model, no vLLM counters | ok | d9191154 | ||||
      | ||||||||||
| 2 | 5/5 | 8/24 | 5 | backend=remote: no local model, no vLLM counters | ok | 160a37f6 | ||||
   | ||||||||||
| 3 | 5/5 | 8/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | f079ab40 |
   not observed (resumed) | ||||||||||
| 4 | 5/5 | 10 9–10/24 | 2 | backend=remote: no local model, no vLLM counters | ok | 63707b06 | ||||
   | ||||||||||
| 5 | 5/5 | 9/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | d8cf3fea |
      not observed (resumed) | ||||||||||
| 6 | 5/5 | 9 8–9/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | df45fbf9 |
      not observed (resumed) | ||||||||||
| 7 | 4/5 | 11 10–11/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | b2a563ed |
     400400not observed (resumed) | ||||||||||
V3-20260910T073546--v3-cc-fable-5-1
claude-code · claude-fable-5-1
| R | gate | cap | min* | req | prompt-tok | gen-tok | ø gen/req | cache* | exit | Build |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | 5/5 | 9/24 | 23 | backend=remote: no local model, no vLLM counters | ok | 2d134391 | ||||
  | ||||||||||
| 2 | 5/5 | 10 9–10/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | 389f6784 |
      not observed (resumed) | ||||||||||
| 3 | 5/5 | 10 9–10/24 | 10 | backend=remote: no local model, no vLLM counters | ok | 6bf4a70f | ||||
      | ||||||||||
| 4 | 5/5 | 13/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | b2584207 |
      not observed (resumed) | ||||||||||
| 5 | 5/5 | 14/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | 67d15f94 |
      not observed (resumed) | ||||||||||
| 6 | 5/5 | 14 14–15/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | d8db22d6 |
      not observed (resumed) | ||||||||||
| 7 | 5/5 | 18 17–18/24 | 0 | — | 0 | 0 | 0 | — | not observed (resumed) | 99858603 |
     not observed (resumed) | ||||||||||
Traces
ein Span je Runde, Projekt antifa-survivors-<zeile>
Span-Attribute je Runde: arena_run_id, arena_config, arena_harness, arena_model, arena_round, arena_exit_code, arena_wall_s, arena_ladder_rung, arena_capability_passed sowie ein arena_<feld> je Modell-Metrik.
* Wallzeit und Prefix-Cache tragen Einschränkungen: die eine misst die gleichzeitige Last mit (2 bis 15 parallele Runden), die andere die Serverwärme und das Alter der Session. Beide stehen hier als Kontext, nicht als Ergebnis.