DeepSeek
DeepSeek-V4-Flash-284B-A13B284B
Benchmark profile and published results.
MoE284B
Position in the field
Best values compared
Best metrics of this model against the minimum, average and maximum of all published systems.
Generation385,9 tok/s
Min 9,4Ø 555,4Max 2.143,6
Unter dem Durchschnitt · 512 Systeme im Feld
Prefill2.421 tok/s
Min 28Ø 6.219Max 21.843
Unter dem Durchschnitt · 512 Systeme im Feld
Time to First Token1.320 ms
Min 59Ø 9.221Max 168.314
Ueber dem Durchschnitt · 512 Systeme im Feld
Performance profile
Throughput by hardware & engine
Each bubble is a GPU, CPU or engine – position shows prompt processing (X) and output speed (Y), bubble size the number of runs.
GPUby graphics card
CPUby processor
Throughput & latency
Performance benchmark
| # | Model / Maker | Metrics | Parallel | GPU / CPU / RAM | Runtime | ||
|---|---|---|---|---|---|---|---|
| 1 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 385,91 tok/s TG Prefill 2.350 · TTFT 30.012 ms | 10× | 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation EditionAMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 2 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 382,08 tok/s TG Prefill 2.421 · TTFT 30.058 ms | 10× | 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation EditionAMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 3 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 206,50 tok/s TG Prefill 2.080 · TTFT 10.489 ms | 5× | 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation EditionAMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 4 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 204,88 tok/s TG Prefill 2.064 · TTFT 10.570 ms | 5× | 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation EditionAMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 5 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 62,33 tok/s TG Prefill 1.735 · TTFT 1.320 ms | 1× | 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation EditionAMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 6 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 62,03 tok/s TG Prefill 1.664 · TTFT 1.377 ms | 1× | 3x NVIDIA RTX PRO 6000 Blackwell Max-Q Workstation EditionAMD Ryzen Threadripper PRO 9965WX 24-Cores · 125 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 7 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 18,80 tok/s TG Prefill 53 · TTFT 260.849 ms | 5× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 8 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 17,96 tok/s TG Prefill 53 · TTFT 258.695 ms | 5× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 9 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 17,47 tok/s TG Prefill 53 · TTFT 259.739 ms | 5× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 10 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 10,87 tok/s TG Prefill 50 · TTFT 231.913 ms | 10× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 11 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 10,76 tok/s TG Prefill 50 · TTFT 232.851 ms | 10× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 12 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 10,35 tok/s TG Prefill 50 · TTFT 233.256 ms | 10× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 13 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 9,95 tok/s TG Prefill 72 · TTFT 174.547 ms | 5× | NVIDIA GeForce RTX 5070 TiAMD Ryzen Threadripper PRO 5975WX 32-Cores · 247 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 14 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 9,13 tok/s TG Prefill 69 · TTFT 186.352 ms | 10× | NVIDIA GeForce RTX 5070 TiAMD Ryzen Threadripper PRO 5975WX 32-Cores · 247 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 15 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 6,49 tok/s TG Prefill 43 · TTFT 53.506 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 16 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 6,46 tok/s TG Prefill 43 · TTFT 53.421 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 17 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 6,41 tok/s TG Prefill 42 · TTFT 53.991 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 18 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 4,29 tok/s TG Prefill 65 · TTFT 35.268 ms | 1× | NVIDIA GeForce RTX 5070 TiAMD Ryzen Threadripper PRO 5975WX 32-Cores · 247 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 19 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 4,27 tok/s TG Prefill 144 · TTFT 14.222 ms | 1× | 3x AMD Radeon AI PRO R9700AMD Ryzen Threadripper PRO 7955WX 16-Cores · 184 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 20 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 3,23 tok/s TG Prefill 28 · TTFT 318.564 ms | 5× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliUD-Q4_K_XL | Details → | |
| 21 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 2,82 tok/s TG Prefill 27 · TTFT 322.266 ms | 10× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliUD-Q4_K_XL | Details → | |
| 22 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 1,20 tok/s TG Prefill 14 · TTFT 148.315 ms | 1× | 2x NVIDIA GeForce RTX 20604x Intel(R) Xeon(R) CPU E5-4620 v2 @ 2.60GHz · 504 GB RAM | llama.cppgodclawUD-Q4_K_XL | Details → | |
| 23 | DeepSeek-V4-Flash-284B-A13B284BDeepSeek Performance benchmarkTimebench 3 - Kombi (Prefill + Generation) | 0,97 tok/s TG Prefill 27 · TTFT 79.443 ms | 1× | NVIDIA GeForce RTX 5090AMD Ryzen 7 5800X3D 8-Core Processor · 126 GB RAM | llama.cppcodex_cliUD-Q4_K_XL | Details → |
Agent & chat rating
Harness benchmark
Noch keine Harnessbenchmarks fuer dieses Modell.
