Contributed byMario AlkaQwen (Alibaba) Qwen3.6-35B-A3B-AWQ
Harness benchmark · measured on 29.09.2026 13:15
Benchmark-ID
run-20260929-112249-dba0f2Decision-Benchmark High V 1.0MoE35.00BRuntime: vllmQuantisierung: AWQ
ℹ
Public note: Decision-Benchmark High V 1.0: identische 200 Aufgaben/10 Familien; eingefrorener Datensatz, unveraenderte Prompts und exakte Bewertung. Korrigierter Qwen-Neulauf: Thinking AKTIV, max_tokens 32768, temperature 1.0,top_p 0.95,top_k 20,min_p 0,presence_penalty 1.5,repetition_penalty 1,seed 20260929. Concurrency 1,Timeout 600s,keine Retries/Tools. AWQ 4-Bit cyankiwi,vLLM 0.27.1. 193/200 richtig; API-Fehler0,Formatfehler4. Mittel18337.799ms; Median15618.341ms; P9540045.952ms. Volle HTTP-Zeit INKLUSIVE Reasoning und erster Anfrage. Separater neuer Benchmark mit Thinking aktiv. Vorheriger Kurzantwortlauf mit deaktiviertem Thinking/32 Tokens (87/200) bleibt unveraendert erhalten. Jev-Vergleich bleibt historischer Cloud-Lauf. Andere Modelle auf ai005 deaktiviert. Unterschiedliches Rechenbudget; kein gleichartiger Latenztest mit Jev.
Points193,00von 200,00
Points rate96,50%
Tasks passed193von 200
Total duration3.667,56s
Ranking in the field
5of 20 systems
Harness benchmark · Primary metric: Score rate (%)
This run is better than 79 % of all comparable systems.
Score rate
97 %
+36 % vs Ø 71,0
Tasks passed
193
+109 % vs Ø 92,5
Total duration
3.667,6 s
+196 % vs Ø 1.240,5
Distribution in the field38 – 100 %
Wie schlägt sich dieser Benchmark mit anderen Modellen?
Qwen3.6-35B-A3B-AWQ same modelNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260929-153935-691df9
100 %
Qwen3.6-35B-A3B-AWQ same modelNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260929-091748-cbe456
100 %
Qwen3.8-27B-Uncensored2× 2x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260929-091747-e13da0
100 %
Open-Jev-27B-v1.1NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260929-091747-5d8477
100 %
Qwen3.6-35B-A3B-AWQ this runNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260929-112249-dba0f2
97 %
Gemma-4-26B-A4B-itNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260722-161611-05d866
91 %
MiniMax-M2.53× 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260722-161611-f6cfe1
81 %
Nemotron-Cascade-2-30B-A3BNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260929-091748-15ef7f
80 %
MiniMax-M2.53× 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260722-161611-54019f
79 %
MiniMax-M2.53× 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260722-161611-a58abe
68 %
MiniMax-M2.53× 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260722-161611-2a2c91
65 %
LayaNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260929-091748-2fc816
61 %
MiniMax-M2.53× 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260722-161612-8d5f5c
61 %
MiniMax-M2.53× 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260722-161611-cfb2d1
58 %
Hardware
GPU: NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · 96 GB VRAM
CPU: AMD Ryzen 9 7945HX with Radeon Graphics
RAM: 60 GB
Mainboard: Shenzhen Meigao Electronic Equipment Co.,Ltd DRFXI (MotherBoard Series)
Setup
Runtime: vllm
Quantization: AWQ
Model: Qwen3.6-35B-A3B-AWQ
Aufgaben (200)
| Task | Status | Points | Duration |
|---|---|---|---|
| 001 - Beschaffung_Optimierung | passed | 1 / 1 | 16,82 s |
| 002 - permission_deny_override | passed | 1 / 1 | 11,85 s |
| 003 - Rechnung_Rundung | passed | 1 / 1 | 12,90 s |
| 004 - Rechnung_Rundung | passed | 1 / 1 | 18,12 s |
| 005 - graph_reachability_blocks | failed | 0 / 1 | 53,52 s |
| 006 - Beschaffung_Optimierung | passed | 1 / 1 | 11,85 s |
| 007 - time_interval_conflicts | passed | 1 / 1 | 15,60 s |
| 008 - time_interval_conflicts | passed | 1 / 1 | 17,98 s |
| 009 - Rechnung_Rundung | passed | 1 / 1 | 18,73 s |
| 010 - permission_deny_override | passed | 1 / 1 | 13,25 s |
| 011 - Abstimmung_Quorum | passed | 1 / 1 | 10,36 s |
| 012 - permission_deny_override | passed | 1 / 1 | 12,77 s |
| 013 - permission_deny_override | passed | 1 / 1 | 13,57 s |
| 014 - budget_dependency_selection | passed | 1 / 1 | 10,90 s |
| 015 - graph_reachability_blocks | passed | 1 / 1 | 16,57 s |
| 016 - Buchungen_Korrekturen | passed | 1 / 1 | 12,92 s |
| 017 - budget_dependency_selection | passed | 1 / 1 | 12,42 s |
| 018 - Lager_Haltbarkeit | passed | 1 / 1 | 15,64 s |
| 019 - version_event_reconstruction | passed | 1 / 1 | 51,96 s |
| 020 - permission_deny_override | passed | 1 / 1 | 12,83 s |
| 021 - Abstimmung_Quorum | passed | 1 / 1 | 9,95 s |
| 022 - Rechnung_Rundung | passed | 1 / 1 | 18,80 s |
| 023 - permission_deny_override | passed | 1 / 1 | 12,50 s |
| 024 - version_event_reconstruction | passed | 1 / 1 | 19,31 s |
| 025 - Lager_Haltbarkeit | failed | 0 / 1 | 26,24 s |
| 026 - time_interval_conflicts | passed | 1 / 1 | 17,50 s |
| 027 - permission_deny_override | passed | 1 / 1 | 15,91 s |
| 028 - Abstimmung_Quorum | passed | 1 / 1 | 9,18 s |
| 029 - Beschaffung_Optimierung | passed | 1 / 1 | 8,73 s |
| 030 - version_event_reconstruction | passed | 1 / 1 | 19,08 s |
| 031 - Lager_Haltbarkeit | passed | 1 / 1 | 11,01 s |
| 032 - Beschaffung_Optimierung | passed | 1 / 1 | 15,08 s |
| 033 - budget_dependency_selection | passed | 1 / 1 | 36,17 s |
| 034 - time_interval_conflicts | passed | 1 / 1 | 21,32 s |
| 035 - time_interval_conflicts | passed | 1 / 1 | 14,95 s |
| 036 - Rechnung_Rundung | passed | 1 / 1 | 19,69 s |
| 037 - budget_dependency_selection | passed | 1 / 1 | 13,00 s |
| 038 - time_interval_conflicts | passed | 1 / 1 | 17,45 s |
| 039 - Buchungen_Korrekturen | passed | 1 / 1 | 16,09 s |
| 040 - version_event_reconstruction | passed | 1 / 1 | 17,72 s |
| 041 - time_interval_conflicts | passed | 1 / 1 | 19,56 s |
| 042 - Lager_Haltbarkeit | passed | 1 / 1 | 15,17 s |
| 043 - Rechnung_Rundung | passed | 1 / 1 | 14,37 s |
| 044 - Rechnung_Rundung | passed | 1 / 1 | 24,54 s |
| 045 - Rechnung_Rundung | passed | 1 / 1 | 16,64 s |
| 046 - time_interval_conflicts | passed | 1 / 1 | 18,95 s |
| 047 - version_event_reconstruction | passed | 1 / 1 | 16,15 s |
| 048 - Lager_Haltbarkeit | passed | 1 / 1 | 15,96 s |
| 049 - Lager_Haltbarkeit | passed | 1 / 1 | 14,53 s |
| 050 - Beschaffung_Optimierung | passed | 1 / 1 | 24,69 s |
| 051 - Rechnung_Rundung | passed | 1 / 1 | 14,16 s |
| 052 - Lager_Haltbarkeit | passed | 1 / 1 | 15,99 s |
| 053 - Buchungen_Korrekturen | passed | 1 / 1 | 11,06 s |
| 054 - budget_dependency_selection | passed | 1 / 1 | 13,91 s |
| 055 - Buchungen_Korrekturen | passed | 1 / 1 | 15,14 s |
| 056 - Abstimmung_Quorum | passed | 1 / 1 | 6,14 s |
| 057 - Buchungen_Korrekturen | passed | 1 / 1 | 11,93 s |
| 058 - Buchungen_Korrekturen | passed | 1 / 1 | 14,70 s |
| 059 - Lager_Haltbarkeit | passed | 1 / 1 | 17,38 s |
| 060 - Rechnung_Rundung | passed | 1 / 1 | 18,73 s |
| 061 - Abstimmung_Quorum | passed | 1 / 1 | 10,33 s |
| 062 - Buchungen_Korrekturen | passed | 1 / 1 | 7,64 s |
| 063 - version_event_reconstruction | passed | 1 / 1 | 18,55 s |
| 064 - Abstimmung_Quorum | passed | 1 / 1 | 11,18 s |
| 065 - permission_deny_override | passed | 1 / 1 | 14,91 s |
| 066 - time_interval_conflicts | passed | 1 / 1 | 14,46 s |
| 067 - Abstimmung_Quorum | passed | 1 / 1 | 9,57 s |
| 068 - Buchungen_Korrekturen | passed | 1 / 1 | 15,05 s |
| 069 - Abstimmung_Quorum | passed | 1 / 1 | 9,71 s |
| 070 - permission_deny_override | passed | 1 / 1 | 14,77 s |
| 071 - permission_deny_override | passed | 1 / 1 | 18,80 s |
| 072 - permission_deny_override | passed | 1 / 1 | 17,50 s |
| 073 - Rechnung_Rundung | passed | 1 / 1 | 17,30 s |
| 074 - version_event_reconstruction | passed | 1 / 1 | 15,75 s |
| 075 - Buchungen_Korrekturen | passed | 1 / 1 | 12,22 s |
| 076 - graph_reachability_blocks | failed | 0 / 1 | 53,36 s |
| 077 - graph_reachability_blocks | passed | 1 / 1 | 21,10 s |
| 078 - time_interval_conflicts | failed | 0 / 1 | 25,26 s |
| 079 - time_interval_conflicts | passed | 1 / 1 | 19,41 s |
| 080 - graph_reachability_blocks | passed | 1 / 1 | 15,02 s |
| 081 - budget_dependency_selection | passed | 1 / 1 | 16,80 s |
| 082 - version_event_reconstruction | passed | 1 / 1 | 20,82 s |
| 083 - Beschaffung_Optimierung | passed | 1 / 1 | 45,49 s |
| 084 - Beschaffung_Optimierung | passed | 1 / 1 | 16,32 s |
| 085 - permission_deny_override | passed | 1 / 1 | 19,51 s |
| 086 - Lager_Haltbarkeit | passed | 1 / 1 | 16,90 s |
| 087 - budget_dependency_selection | passed | 1 / 1 | 47,30 s |
| 088 - Beschaffung_Optimierung | passed | 1 / 1 | 16,90 s |
| 089 - Rechnung_Rundung | passed | 1 / 1 | 19,54 s |
| 090 - Abstimmung_Quorum | passed | 1 / 1 | 9,72 s |
| 091 - Rechnung_Rundung | passed | 1 / 1 | 14,81 s |
| 092 - graph_reachability_blocks | passed | 1 / 1 | 47,71 s |
| 093 - Buchungen_Korrekturen | passed | 1 / 1 | 13,36 s |
| 094 - Buchungen_Korrekturen | passed | 1 / 1 | 11,78 s |
| 095 - version_event_reconstruction | passed | 1 / 1 | 23,42 s |
| 096 - Buchungen_Korrekturen | passed | 1 / 1 | 12,86 s |
| 097 - graph_reachability_blocks | passed | 1 / 1 | 19,55 s |
| 098 - Buchungen_Korrekturen | passed | 1 / 1 | 11,02 s |
| 099 - budget_dependency_selection | passed | 1 / 1 | 13,00 s |
| 100 - Abstimmung_Quorum | passed | 1 / 1 | 10,13 s |
| 101 - permission_deny_override | passed | 1 / 1 | 14,85 s |
| 102 - Beschaffung_Optimierung | passed | 1 / 1 | 20,83 s |
| 103 - Lager_Haltbarkeit | passed | 1 / 1 | 19,72 s |
| 104 - Lager_Haltbarkeit | passed | 1 / 1 | 18,81 s |
| 105 - time_interval_conflicts | passed | 1 / 1 | 13,14 s |
| 106 - graph_reachability_blocks | passed | 1 / 1 | 80,84 s |
| 107 - version_event_reconstruction | passed | 1 / 1 | 20,76 s |
| 108 - Abstimmung_Quorum | passed | 1 / 1 | 11,23 s |
| 109 - permission_deny_override | passed | 1 / 1 | 18,92 s |
| 110 - Abstimmung_Quorum | passed | 1 / 1 | 10,80 s |
| 111 - Abstimmung_Quorum | passed | 1 / 1 | 10,25 s |
| 112 - Lager_Haltbarkeit | passed | 1 / 1 | 15,23 s |
| 113 - permission_deny_override | passed | 1 / 1 | 19,35 s |
| 114 - version_event_reconstruction | passed | 1 / 1 | 13,85 s |
| 115 - version_event_reconstruction | passed | 1 / 1 | 19,71 s |
| 116 - budget_dependency_selection | passed | 1 / 1 | 15,04 s |
| 117 - Rechnung_Rundung | passed | 1 / 1 | 17,78 s |
| 118 - Buchungen_Korrekturen | passed | 1 / 1 | 11,86 s |
| 119 - Rechnung_Rundung | passed | 1 / 1 | 16,73 s |
| 120 - graph_reachability_blocks | passed | 1 / 1 | 10,96 s |
| 121 - version_event_reconstruction | passed | 1 / 1 | 18,92 s |
| 122 - graph_reachability_blocks | passed | 1 / 1 | 29,29 s |
| 123 - graph_reachability_blocks | failed | 0 / 1 | 56,52 s |
| 124 - permission_deny_override | passed | 1 / 1 | 13,79 s |
| 125 - Lager_Haltbarkeit | passed | 1 / 1 | 11,22 s |
| 126 - budget_dependency_selection | passed | 1 / 1 | 14,17 s |
| 127 - Lager_Haltbarkeit | passed | 1 / 1 | 20,73 s |
| 128 - Rechnung_Rundung | passed | 1 / 1 | 15,33 s |
| 129 - budget_dependency_selection | passed | 1 / 1 | 17,26 s |
| 130 - time_interval_conflicts | failed | 0 / 1 | 13,46 s |
| 131 - Abstimmung_Quorum | passed | 1 / 1 | 9,19 s |
| 132 - Buchungen_Korrekturen | passed | 1 / 1 | 12,15 s |
| 133 - budget_dependency_selection | passed | 1 / 1 | 15,56 s |
| 134 - Buchungen_Korrekturen | passed | 1 / 1 | 11,32 s |
| 135 - time_interval_conflicts | passed | 1 / 1 | 21,87 s |
| 136 - time_interval_conflicts | passed | 1 / 1 | 17,59 s |
| 137 - Beschaffung_Optimierung | passed | 1 / 1 | 16,36 s |
| 138 - time_interval_conflicts | passed | 1 / 1 | 24,65 s |
| 139 - Beschaffung_Optimierung | passed | 1 / 1 | 17,73 s |
| 140 - time_interval_conflicts | passed | 1 / 1 | 16,04 s |
| 141 - version_event_reconstruction | passed | 1 / 1 | 20,40 s |
| 142 - graph_reachability_blocks | passed | 1 / 1 | 88,56 s |
| 143 - budget_dependency_selection | passed | 1 / 1 | 31,28 s |
| 144 - Buchungen_Korrekturen | passed | 1 / 1 | 12,53 s |
| 145 - version_event_reconstruction | passed | 1 / 1 | 18,39 s |
| 146 - Buchungen_Korrekturen | passed | 1 / 1 | 13,12 s |
| 147 - graph_reachability_blocks | passed | 1 / 1 | 13,99 s |
| 148 - time_interval_conflicts | passed | 1 / 1 | 23,33 s |
| 149 - budget_dependency_selection | passed | 1 / 1 | 13,37 s |
| 150 - Rechnung_Rundung | passed | 1 / 1 | 17,47 s |
| 151 - permission_deny_override | passed | 1 / 1 | 12,64 s |
| 152 - version_event_reconstruction | passed | 1 / 1 | 13,80 s |
| 153 - Rechnung_Rundung | passed | 1 / 1 | 19,98 s |
| 154 - Beschaffung_Optimierung | passed | 1 / 1 | 10,82 s |
| 155 - version_event_reconstruction | passed | 1 / 1 | 15,35 s |
| 156 - Abstimmung_Quorum | passed | 1 / 1 | 5,27 s |
| 157 - Abstimmung_Quorum | passed | 1 / 1 | 12,04 s |
| 158 - Lager_Haltbarkeit | passed | 1 / 1 | 12,50 s |
| 159 - budget_dependency_selection | passed | 1 / 1 | 16,30 s |
| 160 - budget_dependency_selection | passed | 1 / 1 | 12,25 s |
| 161 - time_interval_conflicts | passed | 1 / 1 | 12,52 s |
| 162 - graph_reachability_blocks | passed | 1 / 1 | 34,27 s |
| 163 - graph_reachability_blocks | passed | 1 / 1 | 12,28 s |
| 164 - Abstimmung_Quorum | passed | 1 / 1 | 9,37 s |
| 165 - budget_dependency_selection | passed | 1 / 1 | 14,78 s |
| 166 - Beschaffung_Optimierung | passed | 1 / 1 | 15,46 s |
| 167 - Buchungen_Korrekturen | passed | 1 / 1 | 12,02 s |
| 168 - Beschaffung_Optimierung | passed | 1 / 1 | 14,61 s |
| 169 - Beschaffung_Optimierung | passed | 1 / 1 | 14,63 s |
| 170 - budget_dependency_selection | passed | 1 / 1 | 13,03 s |
| 171 - Lager_Haltbarkeit | passed | 1 / 1 | 11,90 s |
| 172 - Rechnung_Rundung | passed | 1 / 1 | 16,51 s |
| 173 - Abstimmung_Quorum | passed | 1 / 1 | 13,39 s |
| 174 - Beschaffung_Optimierung | passed | 1 / 1 | 84,22 s |
| 175 - Beschaffung_Optimierung | passed | 1 / 1 | 40,05 s |
| 176 - version_event_reconstruction | passed | 1 / 1 | 21,47 s |
| 177 - Beschaffung_Optimierung | passed | 1 / 1 | 17,86 s |
| 178 - Lager_Haltbarkeit | passed | 1 / 1 | 18,43 s |
| 179 - permission_deny_override | passed | 1 / 1 | 17,86 s |
| 180 - permission_deny_override | passed | 1 / 1 | 19,51 s |
| 181 - version_event_reconstruction | passed | 1 / 1 | 20,22 s |
| 182 - graph_reachability_blocks | passed | 1 / 1 | 16,81 s |
| 183 - Lager_Haltbarkeit | passed | 1 / 1 | 9,72 s |
| 184 - time_interval_conflicts | passed | 1 / 1 | 22,50 s |
| 185 - Abstimmung_Quorum | passed | 1 / 1 | 7,14 s |
| 186 - Rechnung_Rundung | passed | 1 / 1 | 23,66 s |
| 187 - Beschaffung_Optimierung | passed | 1 / 1 | 18,74 s |
| 188 - Buchungen_Korrekturen | passed | 1 / 1 | 11,74 s |
| 189 - graph_reachability_blocks | passed | 1 / 1 | 9,45 s |
| 190 - Abstimmung_Quorum | passed | 1 / 1 | 11,71 s |
| 191 - budget_dependency_selection | passed | 1 / 1 | 13,76 s |
| 192 - permission_deny_override | passed | 1 / 1 | 15,52 s |
| 193 - graph_reachability_blocks | passed | 1 / 1 | 29,74 s |
| 194 - budget_dependency_selection | passed | 1 / 1 | 27,04 s |
| 195 - Lager_Haltbarkeit | passed | 1 / 1 | 15,25 s |
| 196 - graph_reachability_blocks | passed | 1 / 1 | 13,01 s |
| 197 - graph_reachability_blocks | failed | 0 / 1 | 17,61 s |
| 198 - Lager_Haltbarkeit | passed | 1 / 1 | 17,79 s |
| 199 - Beschaffung_Optimierung | passed | 1 / 1 | 27,00 s |
| 200 - version_event_reconstruction | passed | 1 / 1 | 19,08 s |
Configuration
# LLM-Benchmark Konfiguration
# Modell : Qwen3.6-35B-A3B-AWQ
# Engine : vLLM
# Run-ID : run-20260929-112249-dba0f2
# GPU : NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition
# CPU : AMD Ryzen 9 7945HX with Radeon Graphics
# RAM : 60 GB
bench@llm-benchmark:~$ vllm serve \
--model Qwen3.6-35B-A3B-AWQ \
temperature 1 \
top_p 0.95 \
top_k 20 \
min_p 0 \
presence_penalty 1.5 \
repetition_penalty 1 \
seed 20260929 \
max_tokens 32768 \
enable_thinking \
model Qwen3.6-35B-A3B-AWQ \
endpoint http://10.220.4.14:8000/v1/chat/completions
| Engine?Die Inferenz-Software, die das Modell ausliefert (z.B. vLLM oder llama.cpp). Sie bestimmt Geschwindigkeit, unterstuetzte Modellformate und welche Parameter ueberhaupt verfuegbar sind. | vllm |
| Modellalias?Der Name, unter dem das Modell ueber die API angesprochen wird. Genau dieser Wert muss im Request-Feld 'model' stehen. | Qwen3.6-35B-A3B-AWQ |
| temperature | 1 |
| top_p | 0.95 |
| top_k | 20 |
| min_p | 0 |
| presence_penalty | 1.5 |
| repetition_penalty | 1 |
| Seed?Zufalls-Seed fuer reproduzierbare Ausgaben. -1 = zufaellig. | 20260929 |
| max_tokens | 32768 |
| enable_thinking | aktiv |
| Modellpfad?Pfad zur GGUF-Modelldatei, die geladen und ausgeliefert wird. | Qwen3.6-35B-A3B-AWQ |
| endpoint | http://10.220.4.14:8000/v1/chat/completions |
| execution_type | local |
Contributed by
Mario Alka Administrator
@marioalka
Ich bin Unternehmer, Softwareentwickler und KI-Enthusiast. Seit vielen Jahren entwickle ich Unternehmenssoftware und beschäftige mich inzwischen fast täglich mit lokalen LLMs, KI-Agenten und leistungsfähiger KI-Hardware.
Mit LLM-Benchmark.de möchte ich eine Plattform schaffen, auf der Modelle, GPUs und Agenten objektiv und reproduzierbar miteinander verglichen werden.