Sail Research
The most efficient inference for long-horizon agents
Inference built for agents that run for hours, pricing latency tolerance as a dial and pairing it with persistent VM sandboxes for long-horizon workloads.
Pricing
Usage based, no flat monthly plan
Usage-based pricing per million tokens with no seat or subscription fee, and $5 of free credits monthly - rates vary by model and completion window, for example GLM-5.2 runs $0.80 input and $3.00 output per million tokens on asap, dropping to $0.40 and $1.80 on flex - cached input is billed separately - enterprise volume pricing is custom
From the vendor pricing page, 2026-08-28
About Sail Research
Tags
Pricing
Usage based, no flat monthly plan
Usage-based pricing per million tokens with no seat or subscription fee, and $5 of free credits monthly - rates vary by model and completion window, for example GLM-5.2 runs $0.80 input and $3.00 output per million tokens on asap, dropping to $0.40 and $1.80 on flex - cached input is billed separately - enterprise volume pricing is custom
From the vendor pricing page, 2026-08-28
