Output Speed (tokens/s)

Name: Output Speed (tokens/s)
Creator: KnowAI Space
License: https://www.gnu.org/licenses/gpl-3.0.html

About This Benchmark

Number of output tokens generated per second via API. Higher values mean faster responses. Median measured by Artificial Analysis.

Source: Artificial Analysis

Rank	Model
#1	Google Gemini 2.5 Flash	213 tok/s
#2	ChatGPT GPT-5.4 Nano	197 tok/s
#3	Google Gemini 3.1 Flash Lite	196 tok/s
#4	Amazon Nova 2 Lite	180 tok/s
#5	Google Gemini 3 Flash	177 tok/s
#6	Grok Grok 4.20	175 tok/s
#7	ChatGPT GPT-5.4 Mini	172 tok/s
#8	Grok Grok 4.20 (Reasoning)	167 tok/s
#9	ChatGPT GPT-5.4	151 tok/s
#10	Mistral AI Mistral Small 4	150 tok/s
#11	ChatGPT GPT-5 Nano	135 tok/s
#12	Meta AI Llama 4 Scout	126 tok/s
#13	Google Gemini 3.1 Pro	124 tok/s
#14	Google Nano Banana	122 tok/s
#15	Meta AI Llama 4 Maverick	116 tok/s
#16	Google Gemini 2.5 Pro	116 tok/s
#17	Meituan Longcat Flash Chat	115 tok/s
#18	Grok Grok 4.1 Fast (Reasoning)	114 tok/s
#19	Google Gemini 2.5 Flash Lite	105 tok/s
#20	Anthropic Claude Haiku 4.5	99 tok/s
#21	Grok Grok 4.1 Fast	98 tok/s
#22	ChatGPT GPT-4.1	93 tok/s
#23	ChatGPT GPT OSS 120B	86 tok/s
#24	Google Nano Banana 2	82 tok/s
#25	NVIDIA Nemotron 3 Super	80 tok/s
#26	ChatGPT GPT-5	77 tok/s
#27	ChatGPT GPT-5 Mini	75 tok/s
#28	Arcee AI Trinity Large Thinking	70 tok/s
#29	MiniMax MiniMax M2.5	68 tok/s
#30	Xiaomi MiMo-V2-Pro	67 tok/s
#31	ChatGPT GPT-4o Mini Transcribe	53 tok/s
#32	Anthropic Claude Opus 4.5	51 tok/s
#33	Z.ai GLM-5	51 tok/s
#34	Qwen Qwen3.5 397B A17B	49 tok/s
#35	Qwen Qwen3.6 Plus	49 tok/s
#36	DeepSeek DeepSeek V3.2	47 tok/s
#37	Anthropic Claude Opus 4.6	45 tok/s
#38	Anthropic Claude Sonnet 4.6	45 tok/s
#39	Anthropic Claude Sonnet 4	45 tok/s
#40	Anthropic Claude Sonnet 4.5	44 tok/s
#41	MiniMax MiniMax M2.7	39 tok/s
#42	Z.ai GLM-5.1	36 tok/s
#43	Anthropic Claude Opus 4	34 tok/s
#44	Anthropic Claude Opus 4.1	34 tok/s
#45	ChatGPT GPT-4o Transcribe	31 tok/s
#46	Moonshot AI Kimi K2.5	30 tok/s
#47	Google Nano Banana Pro	28 tok/s
#48	Baidu ERNIE 4.5 300B A47B	25 tok/s
#49	Google Gemma 4 31B	14 tok/s
#50	ChatGPT GPT-5.4 Pro	6 tok/s