Contributed byMario AlkaAleph Alpha Kolibri-1
Harness benchmark · measured on 08.10.2026 07:58
Benchmark-ID
run-20261008-125923-43651dKolibri-Qwen Diverse Deterministic 310 V2.0MoE78.1BRuntime: deterministic310Quantisierung: FP8
⚠
Configuration note: Es sind 2 Grafikkarten installiert, aber nur 1 wurde fuer diesen Lauf genutzt. Mit allen 2 GPUs koennte der Durchsatz deutlich hoeher liegen.
Points289,00von 310,00
Points rate93,23%
Tasks passed289von 310
Total duration2.504,67s
Ranking in the field
8of 27 systems
Harness benchmark · Primary metric: Score rate (%)
This run is better than 73 % of all comparable systems.
Score rate
93 %
+30 % vs Ø 71,5
Tasks passed
289
+154 % vs Ø 113,7
Total duration
2.504,7 s
+114 % vs Ø 1.169,9
Distribution in the field37 – 100 %
Wie schlägt sich dieser Benchmark mit anderen Modellen?
Kolibri-1 same modelNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20261008-165122-c5dea7
100 %
Qwen3.6-35B-A3B-AWQNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260929-153935-691df9
100 %
Qwen3.6-35B-A3B-AWQNVIDIA 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 %
Kolibri-1 same modelNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20261008-125923-3f1672
99 %
Qwen3.6-35B-A3B-AWQNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20260929-112249-dba0f2
97 %
Kolibri-1 this runNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20261008-125923-43651d
93 %
Kolibri-1 same modelNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition · run-20261008-125923-5b5304
91 %
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 %
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: deterministic310
Quantization: FP8
Model: Kolibri-1
Aufgaben (310)
| Task | Status | Points | Duration |
|---|---|---|---|
| arithmetic-01 | passed | 1 / 1 | 2,70 s |
| arithmetic-02 | passed | 1 / 1 | 4,23 s |
| arithmetic-03 | passed | 1 / 1 | 5,73 s |
| arithmetic-04 | passed | 1 / 1 | 2,90 s |
| arithmetic-05 | passed | 1 / 1 | 1,97 s |
| arithmetic-06 | passed | 1 / 1 | 2,51 s |
| arithmetic-07 | passed | 1 / 1 | 2,83 s |
| arithmetic-08 | passed | 1 / 1 | 6,10 s |
| arithmetic-09 | passed | 1 / 1 | 3,03 s |
| arithmetic-10 | passed | 1 / 1 | 1,61 s |
| arithmetic-11 | passed | 1 / 1 | 2,67 s |
| arithmetic-12 | passed | 1 / 1 | 3,18 s |
| arithmetic-13 | passed | 1 / 1 | 8,30 s |
| arithmetic-14 | passed | 1 / 1 | 3,58 s |
| arithmetic-15 | passed | 1 / 1 | 1,84 s |
| ordering_logic-01 | failed | 0 / 1 | 5,79 s |
| ordering_logic-02 | passed | 1 / 1 | 4,36 s |
| ordering_logic-03 | passed | 1 / 1 | 5,23 s |
| ordering_logic-04 | passed | 1 / 1 | 4,79 s |
| ordering_logic-05 | passed | 1 / 1 | 3,57 s |
| ordering_logic-06 | passed | 1 / 1 | 5,11 s |
| ordering_logic-07 | passed | 1 / 1 | 8,15 s |
| ordering_logic-08 | passed | 1 / 1 | 5,89 s |
| ordering_logic-09 | passed | 1 / 1 | 7,75 s |
| ordering_logic-10 | passed | 1 / 1 | 6,27 s |
| ordering_logic-11 | failed | 0 / 1 | 4,35 s |
| ordering_logic-12 | passed | 1 / 1 | 6,94 s |
| ordering_logic-13 | passed | 1 / 1 | 5,54 s |
| ordering_logic-14 | passed | 1 / 1 | 6,78 s |
| ordering_logic-15 | passed | 1 / 1 | 4,84 s |
| extraction-01 | passed | 1 / 1 | 6,39 s |
| extraction-02 | passed | 1 / 1 | 5,27 s |
| extraction-03 | passed | 1 / 1 | 6,08 s |
| extraction-04 | passed | 1 / 1 | 5,02 s |
| extraction-05 | passed | 1 / 1 | 4,49 s |
| extraction-06 | passed | 1 / 1 | 5,27 s |
| extraction-07 | passed | 1 / 1 | 3,71 s |
| extraction-08 | passed | 1 / 1 | 5,17 s |
| extraction-09 | passed | 1 / 1 | 8,16 s |
| extraction-10 | passed | 1 / 1 | 5,02 s |
| extraction-11 | passed | 1 / 1 | 7,02 s |
| extraction-12 | passed | 1 / 1 | 7,68 s |
| extraction-13 | passed | 1 / 1 | 6,62 s |
| extraction-14 | passed | 1 / 1 | 4,23 s |
| extraction-15 | passed | 1 / 1 | 9,58 s |
| code_comprehension-01 | passed | 1 / 1 | 2,63 s |
| code_comprehension-02 | passed | 1 / 1 | 8,40 s |
| code_comprehension-03 | passed | 1 / 1 | 15,60 s |
| code_comprehension-04 | passed | 1 / 1 | 8,40 s |
| code_comprehension-05 | passed | 1 / 1 | 3,35 s |
| code_comprehension-06 | passed | 1 / 1 | 3,39 s |
| code_comprehension-07 | passed | 1 / 1 | 6,68 s |
| code_comprehension-08 | passed | 1 / 1 | 13,58 s |
| code_comprehension-09 | passed | 1 / 1 | 11,93 s |
| code_comprehension-10 | passed | 1 / 1 | 2,68 s |
| code_comprehension-11 | passed | 1 / 1 | 3,78 s |
| code_comprehension-12 | passed | 1 / 1 | 4,63 s |
| code_comprehension-13 | passed | 1 / 1 | 15,81 s |
| code_comprehension-14 | passed | 1 / 1 | 9,60 s |
| code_comprehension-15 | passed | 1 / 1 | 4,90 s |
| grounded_qa-01 | passed | 1 / 1 | 2,57 s |
| grounded_qa-02 | passed | 1 / 1 | 2,73 s |
| grounded_qa-03 | passed | 1 / 1 | 3,30 s |
| grounded_qa-04 | passed | 1 / 1 | 3,28 s |
| grounded_qa-05 | passed | 1 / 1 | 2,22 s |
| grounded_qa-06 | passed | 1 / 1 | 3,93 s |
| grounded_qa-07 | passed | 1 / 1 | 3,27 s |
| grounded_qa-08 | passed | 1 / 1 | 2,11 s |
| grounded_qa-09 | passed | 1 / 1 | 2,42 s |
| grounded_qa-10 | passed | 1 / 1 | 2,79 s |
| grounded_qa-11 | passed | 1 / 1 | 2,25 s |
| grounded_qa-12 | passed | 1 / 1 | 3,40 s |
| grounded_qa-13 | passed | 1 / 1 | 3,11 s |
| grounded_qa-14 | passed | 1 / 1 | 3,04 s |
| grounded_qa-15 | passed | 1 / 1 | 2,69 s |
| classification-01 | passed | 1 / 1 | 5,02 s |
| classification-02 | passed | 1 / 1 | 6,28 s |
| classification-03 | passed | 1 / 1 | 3,26 s |
| classification-04 | passed | 1 / 1 | 3,31 s |
| classification-05 | passed | 1 / 1 | 5,32 s |
| classification-06 | passed | 1 / 1 | 6,17 s |
| classification-07 | passed | 1 / 1 | 3,45 s |
| classification-08 | passed | 1 / 1 | 5,16 s |
| classification-09 | passed | 1 / 1 | 4,48 s |
| classification-10 | passed | 1 / 1 | 3,27 s |
| classification-11 | passed | 1 / 1 | 3,42 s |
| classification-12 | passed | 1 / 1 | 3,78 s |
| classification-13 | passed | 1 / 1 | 4,50 s |
| classification-14 | passed | 1 / 1 | 5,28 s |
| classification-15 | passed | 1 / 1 | 4,89 s |
| path_planning-01 | passed | 1 / 1 | 11,13 s |
| path_planning-02 | passed | 1 / 1 | 11,32 s |
| path_planning-03 | passed | 1 / 1 | 8,81 s |
| path_planning-04 | passed | 1 / 1 | 7,62 s |
| path_planning-05 | passed | 1 / 1 | 7,76 s |
| path_planning-06 | passed | 1 / 1 | 8,91 s |
| path_planning-07 | passed | 1 / 1 | 6,25 s |
| path_planning-08 | passed | 1 / 1 | 9,70 s |
| path_planning-09 | passed | 1 / 1 | 7,02 s |
| path_planning-10 | passed | 1 / 1 | 9,73 s |
| path_planning-11 | passed | 1 / 1 | 6,75 s |
| path_planning-12 | passed | 1 / 1 | 7,94 s |
| path_planning-13 | passed | 1 / 1 | 7,34 s |
| path_planning-14 | passed | 1 / 1 | 8,71 s |
| path_planning-15 | passed | 1 / 1 | 10,77 s |
| relational_query-01 | passed | 1 / 1 | 3,39 s |
| relational_query-02 | passed | 1 / 1 | 2,72 s |
| relational_query-03 | passed | 1 / 1 | 2,56 s |
| relational_query-04 | passed | 1 / 1 | 2,76 s |
| relational_query-05 | passed | 1 / 1 | 2,76 s |
| relational_query-06 | passed | 1 / 1 | 3,66 s |
| relational_query-07 | passed | 1 / 1 | 2,55 s |
| relational_query-08 | passed | 1 / 1 | 2,42 s |
| relational_query-09 | passed | 1 / 1 | 2,49 s |
| relational_query-10 | passed | 1 / 1 | 2,54 s |
| relational_query-11 | passed | 1 / 1 | 4,46 s |
| relational_query-12 | passed | 1 / 1 | 2,53 s |
| relational_query-13 | passed | 1 / 1 | 2,71 s |
| relational_query-14 | passed | 1 / 1 | 2,50 s |
| relational_query-15 | passed | 1 / 1 | 3,47 s |
| date_time-01 | passed | 1 / 1 | 3,98 s |
| date_time-02 | passed | 1 / 1 | 3,85 s |
| date_time-03 | passed | 1 / 1 | 20,98 s |
| date_time-04 | passed | 1 / 1 | 6,25 s |
| date_time-05 | passed | 1 / 1 | 2,88 s |
| date_time-06 | passed | 1 / 1 | 6,72 s |
| date_time-07 | passed | 1 / 1 | 8,53 s |
| date_time-08 | passed | 1 / 1 | 3,43 s |
| date_time-09 | passed | 1 / 1 | 26,02 s |
| date_time-10 | passed | 1 / 1 | 6,49 s |
| date_time-11 | passed | 1 / 1 | 6,88 s |
| date_time-12 | passed | 1 / 1 | 32,42 s |
| date_time-13 | passed | 1 / 1 | 6,26 s |
| date_time-14 | passed | 1 / 1 | 3,50 s |
| date_time-15 | passed | 1 / 1 | 30,19 s |
| units-01 | passed | 1 / 1 | 2,65 s |
| units-02 | passed | 1 / 1 | 3,04 s |
| units-03 | passed | 1 / 1 | 5,35 s |
| units-04 | passed | 1 / 1 | 2,66 s |
| units-05 | passed | 1 / 1 | 2,83 s |
| units-06 | passed | 1 / 1 | 2,95 s |
| units-07 | passed | 1 / 1 | 2,08 s |
| units-08 | passed | 1 / 1 | 3,72 s |
| units-09 | passed | 1 / 1 | 2,52 s |
| units-10 | passed | 1 / 1 | 1,94 s |
| units-11 | passed | 1 / 1 | 2,18 s |
| units-12 | passed | 1 / 1 | 3,67 s |
| units-13 | passed | 1 / 1 | 2,24 s |
| units-14 | passed | 1 / 1 | 2,56 s |
| units-15 | passed | 1 / 1 | 2,37 s |
| string_processing-01 | passed | 1 / 1 | 5,58 s |
| string_processing-02 | passed | 1 / 1 | 2,41 s |
| string_processing-03 | passed | 1 / 1 | 2,68 s |
| string_processing-04 | passed | 1 / 1 | 3,58 s |
| string_processing-05 | passed | 1 / 1 | 2,26 s |
| string_processing-06 | passed | 1 / 1 | 2,69 s |
| string_processing-07 | passed | 1 / 1 | 3,70 s |
| string_processing-08 | passed | 1 / 1 | 2,24 s |
| string_processing-09 | passed | 1 / 1 | 2,35 s |
| string_processing-10 | passed | 1 / 1 | 3,87 s |
| string_processing-11 | passed | 1 / 1 | 2,28 s |
| string_processing-12 | passed | 1 / 1 | 2,55 s |
| string_processing-13 | passed | 1 / 1 | 5,05 s |
| string_processing-14 | passed | 1 / 1 | 3,20 s |
| string_processing-15 | passed | 1 / 1 | 3,12 s |
| sets-01 | passed | 1 / 1 | 3,32 s |
| sets-02 | passed | 1 / 1 | 6,42 s |
| sets-03 | passed | 1 / 1 | 4,21 s |
| sets-04 | passed | 1 / 1 | 3,30 s |
| sets-05 | passed | 1 / 1 | 2,38 s |
| sets-06 | passed | 1 / 1 | 7,61 s |
| sets-07 | passed | 1 / 1 | 2,47 s |
| sets-08 | passed | 1 / 1 | 8,26 s |
| sets-09 | passed | 1 / 1 | 6,89 s |
| sets-10 | passed | 1 / 1 | 2,64 s |
| sets-11 | passed | 1 / 1 | 6,44 s |
| sets-12 | passed | 1 / 1 | 6,35 s |
| sets-13 | passed | 1 / 1 | 2,45 s |
| sets-14 | passed | 1 / 1 | 4,90 s |
| sets-15 | passed | 1 / 1 | 9,90 s |
| constraint_selection-01 | passed | 1 / 1 | 3,85 s |
| constraint_selection-02 | passed | 1 / 1 | 4,10 s |
| constraint_selection-03 | passed | 1 / 1 | 3,78 s |
| constraint_selection-04 | passed | 1 / 1 | 3,74 s |
| constraint_selection-05 | passed | 1 / 1 | 3,63 s |
| constraint_selection-06 | passed | 1 / 1 | 3,83 s |
| constraint_selection-07 | passed | 1 / 1 | 3,73 s |
| constraint_selection-08 | passed | 1 / 1 | 3,46 s |
| constraint_selection-09 | passed | 1 / 1 | 3,63 s |
| constraint_selection-10 | passed | 1 / 1 | 3,73 s |
| constraint_selection-11 | passed | 1 / 1 | 2,64 s |
| constraint_selection-12 | passed | 1 / 1 | 4,10 s |
| constraint_selection-13 | passed | 1 / 1 | 3,89 s |
| constraint_selection-14 | passed | 1 / 1 | 3,66 s |
| constraint_selection-15 | passed | 1 / 1 | 3,03 s |
| probability-01 | passed | 1 / 1 | 1,26 s |
| probability-02 | passed | 1 / 1 | 1,32 s |
| probability-03 | passed | 1 / 1 | 2,81 s |
| probability-04 | passed | 1 / 1 | 1,73 s |
| probability-05 | passed | 1 / 1 | 1,43 s |
| probability-06 | passed | 1 / 1 | 3,17 s |
| probability-07 | passed | 1 / 1 | 2,50 s |
| probability-08 | passed | 1 / 1 | 1,68 s |
| probability-09 | passed | 1 / 1 | 2,99 s |
| probability-10 | passed | 1 / 1 | 2,28 s |
| probability-11 | passed | 1 / 1 | 1,78 s |
| probability-12 | passed | 1 / 1 | 2,99 s |
| probability-13 | passed | 1 / 1 | 2,66 s |
| probability-14 | passed | 1 / 1 | 2,24 s |
| probability-15 | passed | 1 / 1 | 3,17 s |
| ext-bit_circuits-01 | passed | 1 / 1 | 3,09 s |
| ext-bit_circuits-02 | passed | 1 / 1 | 29,82 s |
| ext-bit_circuits-03 | passed | 1 / 1 | 3,31 s |
| ext-bit_circuits-04 | passed | 1 / 1 | 17,08 s |
| ext-bit_circuits-05 | passed | 1 / 1 | 3,21 s |
| ext-bit_circuits-06 | passed | 1 / 1 | 30,60 s |
| ext-bit_circuits-07 | passed | 1 / 1 | 4,57 s |
| ext-bit_circuits-08 | passed | 1 / 1 | 27,41 s |
| ext-bit_circuits-09 | passed | 1 / 1 | 10,75 s |
| ext-bit_circuits-10 | passed | 1 / 1 | 31,39 s |
| ext-weighted_scheduling-01 | passed | 1 / 1 | 24,74 s |
| ext-weighted_scheduling-02 | passed | 1 / 1 | 16,27 s |
| ext-weighted_scheduling-03 | passed | 1 / 1 | 12,24 s |
| ext-weighted_scheduling-04 | passed | 1 / 1 | 18,19 s |
| ext-weighted_scheduling-05 | passed | 1 / 1 | 12,31 s |
| ext-weighted_scheduling-06 | passed | 1 / 1 | 24,36 s |
| ext-weighted_scheduling-07 | passed | 1 / 1 | 11,04 s |
| ext-weighted_scheduling-08 | passed | 1 / 1 | 33,90 s |
| ext-weighted_scheduling-09 | passed | 1 / 1 | 12,55 s |
| ext-weighted_scheduling-10 | passed | 1 / 1 | 8,11 s |
| ext-finite_automata-01 | passed | 1 / 1 | 4,17 s |
| ext-finite_automata-02 | passed | 1 / 1 | 8,55 s |
| ext-finite_automata-03 | passed | 1 / 1 | 2,35 s |
| ext-finite_automata-04 | passed | 1 / 1 | 6,39 s |
| ext-finite_automata-05 | passed | 1 / 1 | 2,00 s |
| ext-finite_automata-06 | passed | 1 / 1 | 10,44 s |
| ext-finite_automata-07 | passed | 1 / 1 | 3,13 s |
| ext-finite_automata-08 | passed | 1 / 1 | 25,34 s |
| ext-finite_automata-09 | passed | 1 / 1 | 4,27 s |
| ext-finite_automata-10 | passed | 1 / 1 | 11,76 s |
| ext-fifo_inventory-01 | passed | 1 / 1 | 3,05 s |
| ext-fifo_inventory-02 | passed | 1 / 1 | 5,38 s |
| ext-fifo_inventory-03 | passed | 1 / 1 | 3,02 s |
| ext-fifo_inventory-04 | passed | 1 / 1 | 2,78 s |
| ext-fifo_inventory-05 | passed | 1 / 1 | 2,81 s |
| ext-fifo_inventory-06 | passed | 1 / 1 | 4,90 s |
| ext-fifo_inventory-07 | passed | 1 / 1 | 3,11 s |
| ext-fifo_inventory-08 | passed | 1 / 1 | 3,08 s |
| ext-fifo_inventory-09 | passed | 1 / 1 | 2,75 s |
| ext-fifo_inventory-10 | passed | 1 / 1 | 5,32 s |
| ext-edit_distance-01 | failed | 0 / 1 | 39,65 s |
| ext-edit_distance-02 | failed | 0 / 1 | 39,71 s |
| ext-edit_distance-03 | failed | 0 / 1 | 39,72 s |
| ext-edit_distance-04 | failed | 0 / 1 | 39,57 s |
| ext-edit_distance-05 | failed | 0 / 1 | 39,68 s |
| ext-edit_distance-06 | failed | 0 / 1 | 39,06 s |
| ext-edit_distance-07 | failed | 0 / 1 | 38,81 s |
| ext-edit_distance-08 | failed | 0 / 1 | 38,79 s |
| ext-edit_distance-09 | failed | 0 / 1 | 38,80 s |
| ext-edit_distance-10 | failed | 0 / 1 | 38,79 s |
| ext-calendar_explicit-01 | passed | 1 / 1 | 13,65 s |
| ext-calendar_explicit-02 | passed | 1 / 1 | 7,29 s |
| ext-calendar_explicit-03 | passed | 1 / 1 | 15,83 s |
| ext-calendar_explicit-04 | passed | 1 / 1 | 14,07 s |
| ext-calendar_explicit-05 | passed | 1 / 1 | 8,97 s |
| ext-calendar_explicit-06 | passed | 1 / 1 | 11,88 s |
| ext-calendar_explicit-07 | passed | 1 / 1 | 17,92 s |
| ext-calendar_explicit-08 | passed | 1 / 1 | 13,42 s |
| ext-calendar_explicit-09 | passed | 1 / 1 | 16,21 s |
| ext-calendar_explicit-10 | passed | 1 / 1 | 18,30 s |
| ext-rolling_checksum-01 | passed | 1 / 1 | 11,03 s |
| ext-rolling_checksum-02 | passed | 1 / 1 | 11,24 s |
| ext-rolling_checksum-03 | passed | 1 / 1 | 14,74 s |
| ext-rolling_checksum-04 | passed | 1 / 1 | 11,07 s |
| ext-rolling_checksum-05 | failed | 0 / 1 | 16,03 s |
| ext-rolling_checksum-06 | passed | 1 / 1 | 10,68 s |
| ext-rolling_checksum-07 | passed | 1 / 1 | 12,16 s |
| ext-rolling_checksum-08 | passed | 1 / 1 | 11,77 s |
| ext-rolling_checksum-09 | passed | 1 / 1 | 11,77 s |
| ext-rolling_checksum-10 | failed | 0 / 1 | 9,31 s |
| ext-grid_path_count-01 | failed | 0 / 1 | 22,48 s |
| ext-grid_path_count-02 | passed | 1 / 1 | 6,36 s |
| ext-grid_path_count-03 | failed | 0 / 1 | 13,73 s |
| ext-grid_path_count-04 | passed | 1 / 1 | 9,46 s |
| ext-grid_path_count-05 | failed | 0 / 1 | 20,62 s |
| ext-grid_path_count-06 | passed | 1 / 1 | 11,47 s |
| ext-grid_path_count-07 | passed | 1 / 1 | 5,79 s |
| ext-grid_path_count-08 | passed | 1 / 1 | 8,12 s |
| ext-grid_path_count-09 | passed | 1 / 1 | 9,93 s |
| ext-grid_path_count-10 | passed | 1 / 1 | 8,16 s |
| ext-sample_variance-01 | failed | 0 / 1 | 7,85 s |
| ext-sample_variance-02 | passed | 1 / 1 | 4,66 s |
| ext-sample_variance-03 | failed | 0 / 1 | 20,73 s |
| ext-sample_variance-04 | passed | 1 / 1 | 7,33 s |
| ext-sample_variance-05 | failed | 0 / 1 | 9,23 s |
| ext-sample_variance-06 | passed | 1 / 1 | 14,39 s |
| ext-sample_variance-07 | passed | 1 / 1 | 3,83 s |
| ext-sample_variance-08 | passed | 1 / 1 | 17,43 s |
| ext-sample_variance-09 | failed | 0 / 1 | 38,37 s |
| ext-sample_variance-10 | passed | 1 / 1 | 11,15 s |
| ext-round_robin-01 | passed | 1 / 1 | 7,96 s |
| ext-round_robin-02 | passed | 1 / 1 | 9,32 s |
| ext-round_robin-03 | passed | 1 / 1 | 4,89 s |
| ext-round_robin-04 | passed | 1 / 1 | 11,99 s |
| ext-round_robin-05 | passed | 1 / 1 | 10,74 s |
| ext-round_robin-06 | passed | 1 / 1 | 4,04 s |
| ext-round_robin-07 | passed | 1 / 1 | 13,78 s |
| ext-round_robin-08 | passed | 1 / 1 | 22,42 s |
| ext-round_robin-09 | passed | 1 / 1 | 4,33 s |
| ext-round_robin-10 | passed | 1 / 1 | 12,18 s |
Configuration
# LLM-Benchmark Konfiguration
# Modell : Kolibri-1
# Engine : vLLM
# Run-ID : run-20261008-125923-43651d
# 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 /model \
--served-model-name Aleph-Alpha/Kolibri-1 \
--max-model-len 32768 \
--max-num-seqs 10 \
--gpu-memory-utilization 0.94 \
--kv-cache-dtype fp8 \
--reasoning-parser kolibri1 \
--tool-call-parser kolibri1 \
--enable-prefix-caching
| 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. | Aleph-Alpha/Kolibri-1 |
| Kontextlaenge?Maximale Anzahl Tokens (Eingabe + erzeugte Ausgabe zusammen), die das Modell pro Anfrage verarbeiten kann. | 32768 |
| Alias?Anzeigename des Modells nach aussen (served model name), unabhaengig vom Dateinamen. | Aleph-Alpha/Kolibri-1 |
| Kontext?Groesse des Kontextfensters in Token. 0 = der beim Training verwendete Kontext des Modells. | 32768 |
| Max-Sequenzen?Maximale Anzahl gleichzeitig bearbeiteter Anfragen (Batch-Groesse). Hoeher = mehr Durchsatz, braucht aber mehr KV-Cache-Speicher. | 10 |
| GPU-Speicher?Anteil des GPU-Speichers (0 bis 1), den vLLM belegen darf. 0.92 = 92 %. Hoeher = mehr Platz fuer den KV-Cache (mehr/laengere parallele Anfragen), aber groesseres Risiko fuer 'Out of Memory'. | 0.94 |
| KV-Cache-Dtype?Datentyp des KV-Cache (auto, fp8). fp8 halbiert etwa den Cache-Speicher -> mehr und laengere parallele Anfragen bei minimalem Qualitaetsverlust. | fp8 |
| Reasoning-Parser?Trennt den Denk-/Reasoning-Teil der Antwort vom eigentlichen Inhalt (z.B. bei Modellen mit <think>-Bloecken). | kolibri1 |
| Tool-Parser?Bestimmt, wie vLLM Tool-Aufrufe aus der Modellantwort ausliest (z.B. hermes, mistral, llama3_json). Muss zum Modell passen. | kolibri1 |
| Prefix-Caching?Speichert bereits berechnete Prompt-Anfaenge und verwendet sie wieder. Beschleunigt Anfragen mit gemeinsamem Praefix (z.B. gleicher System-Prompt) deutlich. | aktiv |
| execution_type | local |
Anzeige
Model comparison
Kolibri-1 on various hardware
All published performance runs of this model – each bubble a variant: position = prefill (X) × generation (Y), bubble size = number of runs. Closer to the top right = faster. ★ Marked gold = this benchmark.
GPUby graphics card
CPUby processor
MBby mainboard
DRVby driver
unbekannt 1.044,4 tok/s
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.