Models
Every model below is one we have actually tested. Flat price per call - no token math. If an upstream provider becomes unstable, we swap it out.
MiniMax-M2.7
MiniMax
Cost-efficient general-purpose model from MiniMax. Good for chat and lightweight tasks.
ctx 205K·openai / anthropic
MiniMax-M2.7-highspeed
MiniMax
High-speed variant of MiniMax M2.7 with lower latency, same capability.
ctx 205K·openai / anthropic
MiniMax-M3
MiniMax
MiniMax flagship model. Stronger reasoning and coding for harder tasks.
ctx 1M·openai / anthropic
glm-5.2
Z.ai
GLM 5.2 from Z.ai. Balanced model for bilingual (Chinese/English) use cases.
ctx 1M·openai / anthropic
kimi-k2.6
Moonshot AI
Kimi K2.6 from Moonshot AI. Long-context model good for document-heavy workflows.
ctx 262K·openai / anthropic