Model Spend Arena2ND ED.
serves endpoints · 2026-10-03

StreamLake

Official site · 🇨🇳 China

Models it serves

15 endpoints, cheapest first. This is the view the model-by-model tables cannot give you: what one provider actually offers, and at what precision.

ModelQuant$/M in$/M outContextUptime 30m
DeepSeek V4 Flashfp8$0.04$0.131,024,00099.9%
GLM-5.3-Flashfp8$0.08$0.281,024,00098.1%
Xiaomi MiMo-V2.5unknown$0.17$0.341,000,00097.5%
DeepSeek V4.1 Flashfp8$0.18$0.721,024,00098.9%
Qwen3 Coder Nextunknown$0.18$0.90256,00098.9%
GLM-5.2fp8$0.28$0.881,024,00099.4%
DeepSeek V3unknown$0.26$1.03128,00099.2%
DeepSeek V3.1 Terminusunknown$0.34$1.03128,00099.1%
MiniMax-M3fp8$0.30$1.201,000,00097.7%
Xiaomi MiMo-V2.5-Prounknown$0.52$1.041,000,000—
DeepSeek V4 Prounknown$0.66$1.981,024,00098.5%
Kimi K2.6fp8$0.60$2.52256,00099.2%
Kimi K2.7 Codeunknown$0.71$3.00256,00098.5%
Qwen3.5 397B A17Bunknown$0.60$3.60256,00097.5%
GLM-5.1fp8$0.97$3.04200,00099.3%