Benchmark Explorer

Explore how models perform on various benchmarks

Benchmarks

🔍

Median Tokens/s

The median number of tokens processed per second, measuring the model's throughput in generating responses.
Source:

Model Performance

#1
GPT OSS 120B
253.0
#2
Gemini 2.5 Flash (Thinking)
235.0
#3
Gemini 3.1 Flash Lite Preview
234.0
#4
Grok 4.20 (Reasoning)
230.0
#5
Grok 4.20
214.0
#6
GPT-5.4 Mini
210.0
#7
Gemini 2.5 Flash
209.0
#8
Grok 3 Mini
191.0
#9
GPT-5.4 Nano
191.0
#10
Devstral Small 2
185.0
#11
Gemini 3.0 Flash
163.0
#12
GPT-5
143.1
#13
GPT-5 Nano (Thinking)
139.0
#14
GPT-5 Nano
135.0
#15
OpenAI o3 Mini (Medium Effort)
131.0
#16
MiMo V2 Flash
127.0
#17
Gemini 2.5 Pro (Thinking)
124.0
#18
Llama 4 Maverick 17B
124.0
#19
GPT-5.1 Codex Max
123.0
#20
OpenAI o3 Mini (High Effort)
123.0
#21
OpenAI o4 Mini (Medium Effort)
122.0
#22
OpenAI o4 Mini (High Effort)
122.0
#23
Llama 2 7B
116.8
#24
GPT-5.2 Codex
116.0
#25
Gemini 3.1 Pro Preview
110.0
#26
Claude Haiku 4.5 (Thinking)
101.0
#27
Claude Haiku 4.5
95.0
#28
GPT-3.5 Turbo
93.6
#29
OpenAI o1
91.0
#30
Llama 3.3 70B
85.0
#31
GPT-4.1
82.0
#32
Grok 3
81.0
#33
GLM-4.7
78.0
#34
Devstral 2
78.0
#35
OpenAI o3 (High Effort)
77.0
#36
OpenAI o3 (Medium Effort)
77.0
#37
Mistral Medium 3.1
76.0
#38
GPT-5 (Thinking)
74.0
#39
GPT-5 Mini
72.0
#40
GPT-5 Mini (Thinking)
69.0
#41
GPT-5.3 Codex
68.0
#42
GPT-5.2
66.0
#43
GPT-5.4
66.0
#44
GPT-4o
64.0
#45
GLM-5
60.0
#46
MiniMax M2.1
57.0
#47
MiniMax M2.5
54.0
#48
Qwen 3
53.0
#49
Claude Opus 4.5 (Thinking)
52.0
#50
Claude 4.0 Sonnet (Thinking)
51.0
#51
MiniMax M2.7
51.0
#52
Grok 4
49.0
#53
Grok 4 (Thinking)
49.0
#54
Claude Sonnet 4.5 (Thinking)
49.0
#55
Mistral Large 3
48.0
#56
Qwen 3 Max Preview
47.0
#57
Claude Opus 4.5
46.0
#58
Qwen 3 Max (Thinking)
46.0
#59
Kimi K2.5
45.0
#60
Claude Opus 4.6 (Thinking)
45.0
#61
Claude Sonnet 4.5
43.0
#62
Claude 4.0 Sonnet
43.0
#63
Claude Sonnet 4.6
43.0
#64
Cohere Command A
43.0
#65
Claude Opus 4.6
41.0
#66
Qwen 3 (Thinking)
40.0
#67
Mistral Large 2
40.0
#68
Kimi K2
39.0
#69
GPT-4o Mini
39.0
#70
Claude 4.1 Opus (Thinking)
35.0
#71
Claude 4.0 Opus (Thinking)
34.0
#72
DeepSeek V3.2 (Thinking)
33.0
#73
DeepSeek V3.2
32.0
#74
Claude 4.0 Opus
32.0
#75
Claude 4.1 Opus
32.0
#76
Llama 3.1 405B
30.0
#77
Phi-4
12.0
#78
Claude 3 Sonnet
0.0
#79
Claude 3.5 Haiku
0.0
#80
Claude 3 Opus
0.0
#81
DeepSeek V3 (Mar 2025)
0.0
#82
DeepSeek V3
0.0
#83
Claude 3.7 Sonnet (Thinking)
0.0
#84
Claude 3.5 Sonnet
0.0
#85
Claude 3.7 Sonnet
0.0
#86
GPT-4.5
0.0
#87
DeepSeek R1
0.0
#88
Cohere Command R+
0.0
#89
Gemini 2.0 Flash (Thinking)
0.0
#90
Gemini 2.0 Pro
0.0
#91
Gemini 2.5 Pro (Jun 2025)
0.0
#92
Gemini 1.5 Pro
0.0
#93
Gemini 2.0 Flash
0.0
#94
Grok 3 (Thinking)
0.0
#95
Gemini 2.0 Flash Thinking (Jan 2025)
0.0
#96
Gemini 2.5 Pro
0.0
#97
Llama 2 13B
0.0
#98
Mistral Large
0.0
#99
Magistral Medium 3.1
0.0
#100
Llama 2 70B
0.0
#101
OpenAI o1 Mini
0.0