Speed Champions Leaderboard

Top 20 models ranked by output speed (tok/s)

Rank	Model	Creator	Output Speed (tok/s)	Price/1M	Speed
🥇	Mercury 2 Inception	Inception	975 tok/s	$0.38	975 tok/s
🥈	Granite 3.3 8B (Non-reasoning) IBM	IBM	445 tok/s	$0.09	445 tok/s
🥉	Gemini 2.5 Flash-Lite Preview (Sep '25) (Reasoning) Google	Google	401 tok/s	$0.17	401 tok/s
4	NVIDIA Nemotron 3 Super 120B A12B (Reasoning) NVIDIA	NVIDIA	386 tok/s	$0.41	386 tok/s
5	Gemini 2.5 Flash-Lite Preview (Sep '25) (Non-reasoning) Google	Google	362 tok/s	$0.17	362 tok/s
6	Granite 4.0 H Small IBM	IBM	330 tok/s	$0.11	330 tok/s
7	Gemini 2.5 Flash-Lite (Reasoning) Google	Google	322 tok/s	$0.17	322 tok/s
8	Ministral 3 3B Mistral	Mistral	293 tok/s	$0.10	293 tok/s
9	Nova Micro Amazon	Amazon	293 tok/s	$0.06	293 tok/s
10	Sarvam 30B (high) Sarvam	Sarvam	283 tok/s	$0.00	283 tok/s
11	gpt-oss-120B (low) OpenAI	OpenAI	273 tok/s	$0.26	273 tok/s
12	gpt-oss-20B (high) OpenAI	OpenAI	272 tok/s	$0.09	272 tok/s
13	gpt-oss-20B (low) OpenAI	OpenAI	266 tok/s	$0.09	266 tok/s
14	gpt-oss-120B (high) OpenAI	OpenAI	262 tok/s	$0.26	262 tok/s
15	Gemini 2.5 Flash-Lite (Non-reasoning) Google	Google	243 tok/s	$0.17	243 tok/s
16	Nova 2.0 Lite (medium) Amazon	Amazon	235 tok/s	$0.85	235 tok/s
17	GPT-5.4 nano (Non-Reasoning) OpenAI	OpenAI	224 tok/s	$0.46	224 tok/s
18	Gemini 2.5 Flash (Reasoning) Google	Google	223 tok/s	$0.85	223 tok/s
19	Qwen3 0.6B (Reasoning) Alibaba	Alibaba	223 tok/s	$0.40	223 tok/s
20	Qwen3 0.6B (Non-reasoning) Alibaba	Alibaba	223 tok/s	$0.19	223 tok/s