Try every model of Prem

Models status

Anthropic
claude-3.5-sonnet-v2
99% 27.94
90 days ago Today
claude-3-haiku
99% 70.85
90 days ago Today
claude-3-sonnet
99% 36.70
90 days ago Today
claude-3.5-haiku
99% 31.90
90 days ago Today
claude-3.5-sonnet
99% 33.19
90 days ago Today
Deepseek
deepseek-r1-distill-llama-70b
98% 181.24
90 days ago Today
deepseek-r1
98% 41.73
90 days ago Today
Gpt
gpt-4o-mini-eu
99% 99.18
90 days ago Today
gpt-4-turbo
99% 30.53
90 days ago Today
gpt-4o-mini
99% 54.61
90 days ago Today
gpt-4o
99% 46.85
90 days ago Today
gpt-4-eu
99% 27.35
90 days ago Today
gpt-4o-eu
99% 88.44
90 days ago Today
Llama
llama-3-8b
99% 421.08
90 days ago Today
llama-3.2-3b
99% 467.62
90 days ago Today
llama-3.1-8b
99% 355.88
90 days ago Today
llama-3-70b
99% 214.26
90 days ago Today
llama-3.3-70b
99% 180.41
90 days ago Today
llama-3.2-1b
99% 531.75
90 days ago Today
llama-3-8b-guard
99% 8.08
90 days ago Today
Mistral
mixtral-8x7b
94% 317.26
90 days ago Today

Models performances

Group
Name
AVG Uptime
AVG t/s
Summarization
Tool use
Hallucination
DeepSeek
deepseek-r1
50%
30.19 t/s
-
Not supported
-
Anthropic
claude-3.5-haiku
50%
16.60 t/s
63%
70%
5%
Anthropic
claude-3.5-sonnet
50%
15.51 t/s
67%
60%
0%
Anthropic
claude-3.5-sonnet-v2
50%
14.35 t/s
68%
60%
5%
Anthropic
claude-3-haiku
50%
35.92 t/s
70%
70%
5%
Anthropic
claude-3-sonnet
50%
21.59 t/s
67%
70%
5%
codellama
prem-codellama-7b
38%
42.62 t/s
-
Not supported
-
gemma2
prem-gemma2-2b
38%
23.69 t/s
-
Not supported
-
gemma2
prem-gemma2-9b
38%
27.68 t/s
-
Not supported
-
gpt
gpt-4-eu
50%
13.90 t/s
59%
72%
15%
gpt
gpt-4o
50%
26.98 t/s
71%
67%
10%
gpt
gpt-4o-eu
50%
48.23 t/s
73%
83%
0%
gpt
gpt-4o-mini
50%
27.70 t/s
74%
83%
5%
gpt
gpt-4o-mini-eu
50%
47.98 t/s
72%
78%
5%
gpt
gpt-4-turbo
50%
11.85 t/s
73%
72%
5%
DeepSeek
deepseek-r1-distill-llama-70b
50%
106.72 t/s
-
Not supported
-
llama
llama-3.1-8b
50%
195.43 t/s
62%
88%
10%
llama
llama-3.2-1b
50%
231.30 t/s
45%
Not supported
20%
llama
llama-3.2-3b
50%
230.62 t/s
57%
Not supported
15%
llama
llama-3.3-70b
50%
112.40 t/s
68%
100%
0%
llama
llama-3-70b
50%
111.02 t/s
66%
85%
5%
llama
llama-3-8b
50%
229.55 t/s
64%
89%
5%
llama
llama-3-8b-guard
50%
3.15 t/s
0%
Not supported
100%
mistral
mixtral-8x7b
0%
0.00 t/s
67%
88%
10%
llama
prem-llama-3.1-8b
25%
20.35 t/s
-
Not supported
-
llama
prem-llama-3.2-1b
25%
10.59 t/s
-
Not supported
-
llama
prem-llama-3.2-3b
25%
32.90 t/s
-
Not supported
-
qwen
prem-qwen2.5-0.5b
38%
24.10 t/s
-
Not supported
-
qwen
prem-qwen2.5-1.5b
38%
18.82 t/s
-
Not supported
-
qwen
prem-qwen2.5-3b
38%
17.53 t/s
-
Not supported
-
qwen
prem-qwen2.5-7b
38%
23.68 t/s
-
Not supported
-