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

Novita

Official site · origin not confirmed

Models it serves

38 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
Ling-3.0-flash-VLbf16$0.02$0.06262,14496.7%
Ling-3.0-flashunknown$0.02$0.06262,144100.0%
Ling-3.0-flash-Finunknown$0.04$0.12262,144100.0%
gpt-oss-20bfp4$0.04$0.15131,072100.0%
Nemotron 3 Nano 30Bfp4$0.05$0.20262,144100.0%
gpt-oss-120bfp4$0.05$0.25131,07299.9%
GLM-5.3-Flashfp8$0.08$0.281,048,57699.1%
Gemma 3 27Bbf16$0.12$0.2098,30499.9%
MiMo-V2.6-Flashfp8$0.14$0.281,048,57699.1%
Gemma 4 26B A4B bf16$0.13$0.40262,14499.7%
Gemma 4 31Bbf16$0.14$0.40262,14476.7%
Xiaomi MiMo-V2.5fp8$0.17$0.341,048,57698.6%
Tencent Hy3unknown$0.14$0.58262,14499.8%
Llama 4 Scoutbf16$0.18$0.59131,07299.9%
Llama 4 Maverickfp8$0.27$0.851,048,57698.9%
DeepSeek V4.1 Flashfp8$0.24$0.961,048,57695.6%
Step 3.7 Flashfp8$0.20$1.15262,14499.8%
MiniMax M2.7fp8$0.27$1.08204,80099.3%
MiniMax-M3fp8$0.30$1.201,000,000100.0%
Qwen3 Coder Nextfp8$0.20$1.50262,144—
MiMo-V2.6-Profp8$0.43$0.871,048,57699.8%
Xiaomi MiMo-V2.5-Prounknown$0.48$0.961,048,57699.3%
DeepSeek V4 Flashfp8$0.41$1.231,048,576100.0%
GLM-5.3fp8$0.42$1.321,048,57699.6%
DeepSeek V4 Flash Visionunknown$0.44$1.321,048,576100.0%
GLM 4.7fp8$0.54$1.98204,800100.0%
GLM 4.6bf16$0.55$2.20204,80099.7%
GLM-5.2fp8$0.65$2.041,048,57699.7%
Qwen3.8-27Bunknown$0.42$3.001,000,00098.8%
Qwen3.5-122B-A10Bbf16$0.40$3.20262,14499.1%
Kimi K2.5unknown$0.57$2.85262,14499.0%
DeepSeek R1fp8$0.70$2.5064,000100.0%
Qwen3.5 397B A17Bunknown$0.60$3.60262,14489.6%
Kimi K2.6unknown$0.80$3.40262,14499.7%
DeepSeek V4 Profp8$0.99$2.971,048,576100.0%
Kimi K2.7 Codeint4$0.91$3.84262,144100.0%
GLM-5.1fp8$1.38$4.40204,80099.0%
Qwen3.8 2.4T A95Bunknown$2.00$6.001,000,000100.0%