The Fastest AI Models: Ranked by Tokens/Sec

Ranked by generation speed (tokens per second). Ideal for high-volume text generation, agents, and automated workflows.

Elite Performers

Frontier Model	LMArena Elo	API Cost (1M)	Throughput
grok-4.20-multi-agent-beta-0309	1476	$0.000002	535.5
o3-mini-high	1363	$0.0000011	229
mercury-2	1346	$2.5e-7	196
grok-4-fast-chat	1421	$2e-7	180
o3-mini	1347	$0.0000011	173
grok-4-1-fast-reasoning	1432	$2e-7	146
o1-2024-12-17	1401	$0.000015	136

*These models represent the Pareto Frontier (optimal cost-to-performance).*