Ollama benchmark Q2 2024 - Exoscale A40 - LLM Performance

Go back to list


Filter:

eval_rate prompt_eval_rate
mean min std mean min std
model
codegemma 73.6 72.8 1.2 222.3 220.9 1.9
codegemma:2b 158.5 140.7 28.4 315.0 261.9 75.2
codellama 97.3 96.5 1.1 268.0 250.9 43.7
codellama:13b 58.2 57.7 0.4 188.9 162.8 45.0
codellama:70b 13.3 13.3 0.0 61.0 61.0 0.1
deepseek-coder 251.6 251.2 0.6 649.0 647.1 2.8
deepseek-coder:33b 26.2 26.2 0.0 75.1 74.8 0.5
deepseek-coder:6.7b 100.0 99.9 0.2 249.7 249.0 0.9
dolphin-mixtral 52.4 52.4 0.0 142.9 142.9 0.0
gemma 77.5 76.5 0.8 222.6 218.2 2.0
gemma:2b 148.4 144.7 3.1 522.8 492.5 14.1
llama2 96.8 95.6 0.7 252.7 243.8 2.9
llama2-uncensored 99.7 97.3 2.4 327.2 325.0 1.3
llama2-uncensored:70b 13.3 13.3 0.0 56.8 56.8 0.0
llama2:13b 58.0 57.4 0.4 162.4 161.4 0.5
llama2:70b 13.3 13.1 0.1 43.5 40.2 1.0
llama3 80.3 78.9 0.7 202.0 191.5 4.7
llama3:70b 12.8 12.8 0.0 39.3 39.2 0.0
llava 95.3 95.1 0.3 223.1 222.7 0.6
llava:13b 58.5 57.8 1.0 170.6 170.6 0.1
llava:34b 25.0 25.0 0.1 79.7 79.1 0.8
mistral 92.0 90.6 0.8 222.8 221.8 0.6
mixtral 53.1 52.4 0.5 148.1 147.4 0.5
orca-mini 156.0 154.7 1.0 477.8 471.1 3.4
orca-mini:13b 57.6 57.4 0.4 184.7 182.8 2.7
orca-mini:70b 13.1 13.1 0.0 50.3 50.3 0.0
orca-mini:7b 95.3 94.9 0.6 289.2 288.7 0.6
phi3 124.1 121.1 1.8 439.8 434.7 3.5
qwen 103.1 102.4 0.9 309.9 304.4 7.8
qwen:0.5b 206.9 205.5 2.0 1042.5 1033.5 12.8
qwen:1.8b 162.4 152.8 8.3 599.4 577.5 19.1
qwen:14b 51.2 51.1 0.1 156.8 155.8 1.5
qwen:32b 24.7 24.6 0.0 73.9 73.9 0.0
qwen:72b 12.5 12.5 0.0 45.5 45.5 0.0
qwen:7b 81.2 81.0 0.3 242.9 242.8 0.0
vicuna 96.8 95.6 1.6 265.3 264.3 0.7
vicuna:13b 57.6 57.4 0.2 170.4 170.3 0.2
vicuna:33b 25.2 25.2 0.1 87.4 87.1 0.4
wizardlm2 91.0 90.7 0.3 233.6 232.2 0.7