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

Phala

Official site · origin not confirmed

Models it serves

18 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
Nemotron 3.5 Lightningunknown$0.07$0.20262,144100.0%
GLM-5.3-Flashfp8$0.12$0.401,048,57699.0%
gpt-oss-120bunknown$0.15$0.60131,072100.0%
Tencent Hy3unknown$0.15$0.64262,14499.7%
DeepSeek V4.1 Flashunknown$0.21$0.841,048,57699.3%
DeepSeek V4 Flashunknown$0.31$0.921,048,576100.0%
Qwen3.6 35B A3Bunknown$0.20$1.27262,144100.0%
Muse Glimmerunknown$0.30$1.10131,072100.0%
Qwen3.8-27Bunknown$0.15$1.881,000,00098.8%
Qwen3.6 27Bunknown$0.32$2.70262,14499.3%
DeepSeek V3.2unknown$1.00$1.00163,84098.3%
Qwen3.5 397B A17Bunknown$0.55$3.50262,14498.7%
GLM-5.3unknown$0.84$2.641,048,576100.0%
DeepSeek V4 Prounknown$0.96$2.881,048,576—
GLM-5.2fp8$1.26$3.001,048,576—
GLM-5.1unknown$1.21$4.20202,75292.1%
Kimi K2.6unknown$1.09$4.60262,14499.6%
Kimi K3unknown$2.25$11.251,048,57699.2%