Created bymario-alka.dePowered bygodcore.denoob2claw.detricoma.de
vL
Modellserver / Inferenz-Engine

vLLM

Alle Performance-Benchmarks, die mit diesem Modellserver gemessen wurden – Durchsatz, Prefill und Latenz.

54Benchmarks
19Modelle
Beste Generation
69,2 tok/s
Qwen3-VL-30B-A3B-Instruct
∅ Generation
31,7 tok/s
Schnitt aus 54 Läufen
Bester Prefill
9.072 tok/s
Prompt-Verarbeitung
Bester TTFT
91 ms
Time to First Token

Top-Modelle auf vLLM

Bester Generierungs-Durchsatz je Modell (tok/s).

MiniMax-M2.5
62,7 tok/s
Qwen3.5-35B-A3B
59,4 tok/s
Qwen3.6-35B-A3B
56,0 tok/s
gpt-oss-20b
52,7 tok/s

Auf welcher Hardware laeuft vLLM am schnellsten?

Bester Durchsatz je Grafikkarte mit diesem Modellserver.

AMD Radeon AI PRO R9700
69,2 tok/s
NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition
62,7 tok/s
NVIDIA GB10
34,4 tok/s
Performanceprofil

Durchsatz mit vLLM

Jede Blase steht für eine CPU, GPU bzw. ein Modell – Position: Prompt-Verarbeitung (X) × Ausgabe (Y), Blasengröße: Anzahl Messläufe.

CPUnach Prozessor

84,266,649,031,413,801.7813.5625.343Prefill (tok/s)Generation (tok/s)AMD Ryzen Threadripper PRO 7955WX 16-Cores - 69,2 tok/s Generation, 4.363 tok/s Prefill, TTFT 799 ms (51 Laufe)AMD Ryzen Threadrippe...AMD Ryzen Threadripper PRO 9965WX 24-Cores - 62,7 tok/s Generation, 368 tok/s Prefill, TTFT 144 ms (1 Lauf)AMD Ryzen Threadrippe...NVIDIA Grace - 34,4 tok/s Generation, 503 tok/s Prefill, TTFT 105 ms (1 Lauf)NVIDIA GraceAMD Ryzen 9 9950X 16-Core Processor - 28,8 tok/s Generation, 599 tok/s Prefill, TTFT 122 ms (1 Lauf)AMD Ryzen 9 9950X 16-...
AMD Ryzen Threadripper PRO 7955WX 16-Cores 69,2 tok/sAMD Ryzen Threadripper PRO 9965WX 24-Cores 62,7 tok/sNVIDIA Grace 34,4 tok/sAMD Ryzen 9 9950X 16-Core Processor 28,8 tok/s

GPUnach Grafikkarte

83,167,451,836,120,501.7743.5485.323Prefill (tok/s)Generation (tok/s)AMD Radeon AI PRO R9700 - 69,2 tok/s Generation, 4.363 tok/s Prefill, TTFT 799 ms (51 Laufe)AMD Radeon AI PRO R97...NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition - 62,7 tok/s Generation, 484 tok/s Prefill, TTFT 133 ms (2 Laufe)NVIDIA RTX PRO 6000 B...NVIDIA GB10 - 34,4 tok/s Generation, 503 tok/s Prefill, TTFT 105 ms (1 Lauf)NVIDIA GB10
AMD Radeon AI PRO R9700 69,2 tok/sNVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition 62,7 tok/sNVIDIA GB10 34,4 tok/s

MODnach Modell

88,166,144,122,00,003.5547.10710.661Prefill (tok/s)Generation (tok/s)Qwen3-VL-30B-A3B-Instruct - 69,2 tok/s Generation, 7.075 tok/s Prefill, TTFT 459 ms (4 Laufe)Qwen3-VL-30B-A3B-Inst...MiniMax-M2.5 - 62,7 tok/s Generation, 368 tok/s Prefill, TTFT 144 ms (1 Lauf)MiniMax-M2.5Qwen3.5-35B-A3B - 59,4 tok/s Generation, 3.889 tok/s Prefill, TTFT 663 ms (4 Laufe)Qwen3.5-35B-A3BQwen3.6-35B-A3B - 56,0 tok/s Generation, 4.086 tok/s Prefill, TTFT 703 ms (3 Laufe)Qwen3.6-35B-A3Bgpt-oss-20b - 52,7 tok/s Generation, 4.971 tok/s Prefill, TTFT 745 ms (3 Laufe)gpt-oss-20bQwen3-30B-A3B-Thinking-2507 - 49,3 tok/s Generation, 4.245 tok/s Prefill, TTFT 1.758 ms (5 Laufe)Qwen3-30B-A3B-Thinkin...Qwen3-Coder-30B-A3B-Instruct - 48,0 tok/s Generation, 4.808 tok/s Prefill, TTFT 898 ms (6 Laufe)Qwen3-Coder-30B-A3B-I...Qwen3-30B-A3B-Instruct-2507 - 45,4 tok/s Generation, 3.444 tok/s Prefill, TTFT 1.209 ms (6 Laufe)Qwen3-30B-A3B-Instruc...Qwen3-Omni-30B-A3B-Thinking - 34,8 tok/s Generation, 5.528 tok/s Prefill, TTFT 767 ms (1 Lauf)Qwen3-Omni-30B-A3B-Th...MiniMax-M2.7-AWQ-4bit - 34,4 tok/s Generation, 503 tok/s Prefill, TTFT 105 ms (1 Lauf)MiniMax-M2.7-AWQ-4bitQwen3.6-27B - 28,8 tok/s Generation, 599 tok/s Prefill, TTFT 122 ms (1 Lauf)Qwen3.6-27BLaguna-XS-2.1-INT4 - 26,9 tok/s Generation, 3.698 tok/s Prefill, TTFT 748 ms (1 Lauf)Laguna-XS-2.1-INT4gemma-4-31b-it-int8 - 24,4 tok/s Generation, 219 tok/s Prefill, TTFT 271 ms (4 Laufe)gemma-4-31b-it-int8Qwen3-Coder-Next - 22,7 tok/s Generation, 2.233 tok/s Prefill, TTFT 1.116 ms (2 Laufe)Qwen3-Coder-Nextgpt-oss-120b - 21,5 tok/s Generation, 4.300 tok/s Prefill, TTFT 556 ms (1 Lauf)gpt-oss-120bMistral-Small-3.1-24B-Instruct-2503 - 15,0 tok/s Generation, 5.364 tok/s Prefill, TTFT 522 ms (3 Laufe)Mistral-Small-3.1-24B...Laguna-S-2.1-INT4 - 13,4 tok/s Generation, 8.629 tok/s Prefill, TTFT 305 ms (1 Lauf)Laguna-S-2.1-INT4Qwen2.5-Coder-32B-Instruct - 9,9 tok/s Generation, 5.318 tok/s Prefill, TTFT 472 ms (6 Laufe)Qwen2.5-Coder-32B-Ins...GLM-4.5-Air - 9,1 tok/s Generation, 5.237 tok/s Prefill, TTFT 407 ms (1 Lauf)GLM-4.5-Air
Qwen3-VL-30B-A3B-Instruct 69,2 tok/sMiniMax-M2.5 62,7 tok/sQwen3.5-35B-A3B 59,4 tok/sQwen3.6-35B-A3B 56,0 tok/sgpt-oss-20b 52,7 tok/sQwen3-30B-A3B-Thinking-2507 49,3 tok/sQwen3-Coder-30B-A3B-Instruct 48,0 tok/sQwen3-30B-A3B-Instruct-2507 45,4 tok/sQwen3-Omni-30B-A3B-Thinking 34,8 tok/sMiniMax-M2.7-AWQ-4bit 34,4 tok/sQwen3.6-27B 28,8 tok/sLaguna-XS-2.1-INT4 26,9 tok/sgemma-4-31b-it-int8 24,4 tok/sQwen3-Coder-Next 22,7 tok/sgpt-oss-120b 21,5 tok/sMistral-Small-3.1-24B-Instruct-2503 15,0 tok/sLaguna-S-2.1-INT4 13,4 tok/sQwen2.5-Coder-32B-Instruct 9,9 tok/sGLM-4.5-Air 9,1 tok/s
Alle Messungen

61 Benchmark-Läufe

Jeder veroeffentlichte Lauf mit vLLM.

← Alle Engines
#Modell / HerstellerTypMesswerteGPU / CPU / RAMRuntimeDatum
1MiniMax-M2.5MiniMaxTool Usage Standard 1.0Harnessbenchmark
81,4 %
2.427/2.980 Pkt · 16/16 Aufg.
3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition34x AMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAMvLLMgodclaw BF1613.07.2026
2Qwen3-VL-30B-A3B-InstructQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
69,19 tok/s
Prefill 7.289 · TTFT 412 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
3MiniMax-M2.5MiniMaxTool Usage Standard 1.0Harnessbenchmark
68,2 %
2.033/2.980 Pkt · 16/16 Aufg.
3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition34x AMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAMvLLMgodclaw BF1613.07.2026
4MiniMax-M2.5MiniMaxPerformancetest Small 1.0Performancebenchmark
62,70 tok/s
Prefill 368 · TTFT 144 ms
3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition34x AMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAMvLLMgodclaw BF1620.07.2026
5MiniMax-M2.7-AWQ-4bitMiniMaxTool-Parcours Dossier 40 V1.0Harnessbenchmark
61,1 %
3.641/5.960 Pkt · 40/40 Aufg.
NVIDIA GB10NVIDIA Grace · 120 GB RAMvLLMgodclaw AWQ19.07.2026
6MiniMax-M2.5MiniMaxTool-Parcours Dossier 40 V1.0Harnessbenchmark
60,8 %
3.622/5.960 Pkt · 34/40 Aufg.
3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition34x AMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAMvLLMgodclaw BF1619.07.2026
7Qwen3.5-35B-A3BQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
59,41 tok/s
Prefill 3.761 · TTFT 569 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
8MiniMax-M2.5MiniMaxTool Usage Standard 1.0Harnessbenchmark
57,5 %
1.714/2.980 Pkt · 16/16 Aufg.
3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition34x AMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAMvLLMgodclaw BF1615.07.2026
9Qwen3.6-35B-A3BQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
55,99 tok/s
Prefill 3.520 · TTFT 622 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
10gpt-oss-20bOpenAITimebench 3 - Kombi (Prefill + Generation)Performancebenchmark
52,72 tok/s
Prefill 4.993 · TTFT 874 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw22.07.2026
11Qwen3-30B-A3B-Thinking-2507Qwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
49,30 tok/s
Prefill 320 · TTFT 6.725 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
12Qwen3-30B-A3B-Thinking-2507Qwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
48,40 tok/s
Prefill 3.509 · TTFT 649 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ22.07.2026
13Qwen3-Coder-30B-A3B-InstructQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
47,99 tok/s
Prefill 867 · TTFT 2.482 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
14Qwen3-Coder-30B-A3B-InstructQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
47,65 tok/s
Prefill 5.913 · TTFT 437 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ22.07.2026
15Qwen3-Coder-30B-A3B-InstructQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
46,52 tok/s
Prefill 4.080 · TTFT 752 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ22.07.2026
16Qwen3-VL-30B-A3B-InstructQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
46,18 tok/s
Prefill 5.211 · TTFT 480 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ22.07.2026
17Qwen3-30B-A3B-Instruct-2507Qwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
45,43 tok/s
Prefill 861 · TTFT 2.500 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
18Qwen3-30B-A3B-Instruct-2507Qwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
44,17 tok/s
Prefill 4.851 · TTFT 489 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ22.07.2026
19Qwen3-30B-A3B-Instruct-2507Qwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
43,75 tok/s
Prefill 913 · TTFT 2.357 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
20Qwen3-30B-A3B-Thinking-2507Qwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
42,04 tok/s
Prefill 3.904 · TTFT 581 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ22.07.2026
21Qwen3-30B-A3B-Thinking-2507Qwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
41,95 tok/s
Prefill 8.044 · TTFT 378 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
22Qwen3-30B-A3B-Thinking-2507Qwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
41,21 tok/s
Prefill 5.446 · TTFT 460 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ22.07.2026
23Qwen3.6-27BQwen (Alibaba)Tool Usage Standard 1.0Harnessbenchmark
40,9 %
1.220/2.980 Pkt · 16/16 Aufg.
NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition32x AMD Ryzen 9 9950X 16-Core Processor · 92 GB RAMvLLMgodclaw15.07.2026
24Qwen3-Coder-30B-A3B-InstructQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
40,50 tok/s
Prefill 5.664 · TTFT 755 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
25Qwen3-Coder-30B-A3B-InstructQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
39,74 tok/s
Prefill 3.947 · TTFT 575 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ22.07.2026
26Qwen3.5-35B-A3BQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
39,70 tok/s
Prefill 4.998 · TTFT 387 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ22.07.2026
27Qwen3-30B-A3B-Instruct-2507Qwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
39,27 tok/s
Prefill 5.443 · TTFT 774 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
28Qwen3-Coder-30B-A3B-InstructQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
38,34 tok/s
Prefill 8.377 · TTFT 387 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
29Qwen3.6-27BQwen (Alibaba)Tool-Parcours Dossier 40 V1.0Harnessbenchmark
38,0 %
2.267/5.960 Pkt · 38/40 Aufg.
NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition32x AMD Ryzen 9 9950X 16-Core Processor · 92 GB RAMvLLMgodclaw19.07.2026
30Qwen3-VL-30B-A3B-InstructQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
36,87 tok/s
Prefill 7.939 · TTFT 473 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
31gpt-oss-20bOpenAITimebench 3 - Kombi (Prefill + Generation)Performancebenchmark
35,66 tok/s
Prefill 6.540 · TTFT 375 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw22.07.2026
32Qwen3-Omni-30B-A3B-ThinkingQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
34,82 tok/s
Prefill 5.528 · TTFT 767 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ22.07.2026
33Qwen3-30B-A3B-Instruct-2507Qwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
34,80 tok/s
Prefill 4.274 · TTFT 570 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ22.07.2026
34MiniMax-M2.7-AWQ-4bitMiniMaxPerformancetest Small 1.0Performancebenchmark
34,37 tok/s
Prefill 503 · TTFT 105 ms
NVIDIA GB10NVIDIA Grace · 120 GB RAMvLLMgodclaw AWQ20.07.2026
35Qwen3-VL-30B-A3B-InstructQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
34,36 tok/s
Prefill 7.861 · TTFT 473 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
36Qwen3.6-35B-A3BQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
34,21 tok/s
Prefill 5.971 · TTFT 356 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ22.07.2026
37Qwen3.6-35B-A3BQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
33,19 tok/s
Prefill 2.767 · TTFT 1.132 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ22.07.2026
38Qwen3-30B-A3B-Instruct-2507Qwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
32,76 tok/s
Prefill 4.322 · TTFT 563 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ22.07.2026
39Qwen3.5-35B-A3BQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
30,28 tok/s
Prefill 1.804 · TTFT 1.281 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
40Qwen3.5-35B-A3BQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
29,06 tok/s
Prefill 4.992 · TTFT 415 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ22.07.2026
41Qwen3.6-27BQwen (Alibaba)Timebench 2 - Token-Generation / Output-DurchsatzPerformancebenchmark
28,80 tok/s
Prefill 599 · TTFT 122 ms
NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation Edition32x AMD Ryzen 9 9950X 16-Core Processor · 92 GB RAMvLLMgodclaw20.07.2026
42gpt-oss-20bOpenAITimebench 3 - Kombi (Prefill + Generation)Performancebenchmark
28,70 tok/s
Prefill 3.379 · TTFT 986 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw22.07.2026
43Laguna-XS-2.1-INT4PoolsideTimebench 3 - Kombi (Prefill + Generation)Performancebenchmark
26,88 tok/s
Prefill 3.698 · TTFT 748 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw INT422.07.2026
44gemma-4-31b-it-int8GooglePerformancetest Small 1.0Performancebenchmark
24,36 tok/s
Prefill 309 · TTFT 91 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw INT820.07.2026
45Qwen3-Coder-NextQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
22,73 tok/s
Prefill 2.230 · TTFT 1.115 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
46Qwen3-Coder-NextQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
22,53 tok/s
Prefill 2.237 · TTFT 1.117 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
47gpt-oss-120bOpenAITimebench 3 - Kombi (Prefill + Generation)Performancebenchmark
21,48 tok/s
Prefill 4.300 · TTFT 556 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw22.07.2026
48Mistral-Small-3.1-24B-Instruct-2503Mistral AITimebench 3 - Kombi (Prefill + Generation)Performancebenchmark
15,00 tok/s
Prefill 9.072 · TTFT 245 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw INT422.07.2026
49Laguna-S-2.1-INT4PoolsideTimebench 3 - Kombi (Prefill + Generation)Performancebenchmark
13,37 tok/s
Prefill 8.629 · TTFT 305 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw INT422.07.2026
50gemma-4-31b-it-int8GooglePerformancetest Small 1.0Performancebenchmark
12,45 tok/s
Prefill 208 · TTFT 135 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw INT820.07.2026
51gemma-4-31b-it-int8GoogleTimebench 2 - Token-Generation / Output-DurchsatzPerformancebenchmark
11,19 tok/s
Prefill 162 · TTFT 716 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw INT820.07.2026
52gemma-4-31b-it-int8GooglePerformancetest Small 1.0Performancebenchmark
10,54 tok/s
Prefill 199 · TTFT 141 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw INT820.07.2026
53Qwen2.5-Coder-32B-InstructQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
9,86 tok/s
Prefill 8.244 · TTFT 264 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
54Qwen2.5-Coder-32B-InstructQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
9,84 tok/s
Prefill 7.325 · TTFT 311 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
55GLM-4.5-AirZ.ai (Zhipu)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
9,08 tok/s
Prefill 5.237 · TTFT 407 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ22.07.2026
56Mistral-Small-3.1-24B-Instruct-2503Mistral AITimebench 3 - Kombi (Prefill + Generation)Performancebenchmark
6,98 tok/s
Prefill 4.076 · TTFT 559 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw INT422.07.2026
57Mistral-Small-3.1-24B-Instruct-2503Mistral AITimebench 3 - Kombi (Prefill + Generation)Performancebenchmark
6,13 tok/s
Prefill 2.944 · TTFT 763 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw INT422.07.2026
58Qwen2.5-Coder-32B-InstructQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
5,57 tok/s
Prefill 4.108 · TTFT 529 ms
AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
59Qwen2.5-Coder-32B-InstructQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
5,54 tok/s
Prefill 5.413 · TTFT 420 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
60Qwen2.5-Coder-32B-InstructQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
5,30 tok/s
Prefill 3.582 · TTFT 607 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026
61Qwen2.5-Coder-32B-InstructQwen (Alibaba)Timebench 3 - Kombi (Prefill + Generation)Performancebenchmark
5,24 tok/s
Prefill 3.234 · TTFT 701 ms
3x AMD Radeon AI PRO R970032x AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAMvLLMgodclaw AWQ21.07.2026