Time to First Answer Token

Artificial Analysis · 50 models

Rank	Model	Score
1	Gemini 2.5 Flash-Lite (Non-reasoning)	0.29
2	Gemini 2.5 Flash (Non-reasoning)	0.38
3	Grok 4.20 0309 v2 (Non-reasoning)	0.5
4	Qwen3.5 9B (Non-reasoning)	0.51
5	Celeris-1	0.54
5	Claude 4.5 Haiku (Non-reasoning)	0.54
5	NVIDIA Nemotron 3 Nano 30B A3B (Non-reasoning)	0.54
8	Qwen3 30B A3B (Non-reasoning)	0.56
9	DeepSeek V3 0324	0.64
10	Grok 4.3 (Non-reasoning)	0.66
11	DeepSeek V3.1 Terminus (Non-reasoning)	0.72
12	GPT-5.4 mini (Non-reasoning)	0.73
13	Gemma 4 E4B (Non-reasoning)	0.75
14	GPT-5.4 nano (Non-reasoning)	0.76
15	GPT-5 nano (Minimal)	0.78
15	Gemini 3.5 Flash (Minimal)	0.78
15	Nova Micro	0.78
18	GPT-4.1 mini	0.8
18	GPT-5.4 (Non-reasoning)	0.8
20	Claude Sonnet 5 (Non-reasoning, High)	0.81
20	GPT-5.6 Luna (Non-reasoning)	0.81
20	Ministral 3 3B	0.81
23	GPT-5.6 Terra (Non-reasoning)	0.82
23	GPT-6 Luna (Non-reasoning)	0.82
25	GPT-5.2 (Non-reasoning)	0.83
25	Llama 4 Scout	0.83
25	Qwen3.5 4B (Non-reasoning)	0.83
28	Gemini 3 Flash Preview (Non-reasoning)	0.84
29	Phi-4 Mini Instruct	0.85
30	GPT-4.1	0.86
30	Mistral Small 4 (Non-reasoning)	0.86
32	GPT-6 Sol (Non-reasoning)	0.87
33	DeepSeek V4.1 Flash (Non-reasoning)	0.88
33	GPT-5 mini (Minimal)	0.88
33	GPT-5.6 Sol (Non-reasoning)	0.88
36	Llama 3.1 Instruct 8B	0.89
36	Mistral Small 3.2	0.89
36	Qwen3 Coder Next	0.89
39	Qwen3 Next 80B A3B Instruct	0.9
40	Nova Lite	0.91
41	DeepSeek V3 (Dec '24)	0.92
41	GPT-4o (Nov '24)	0.92
41	Gemma 4 26B A4B (Non-reasoning)	0.92
44	GPT-6.1 Terra (Non-reasoning)	0.94
44	Mistral Small 3.1	0.94
44	Qwen3 30B A3B 2507 Instruct	0.94
47	GPT-4o (Aug '24)	0.95
47	GPT-4o mini	0.95
47	GPT-5.5 (Non-reasoning)	0.95
47	Qwen3 Coder 30B A3B Instruct	0.95