One price per model.
Always the fastest node.
Every model below routes across multiple providers. You pay the published rate no matter which one serves you, no platform fee.
Not writing code? Run these models in Claude Desktop or Codex.
deepseek-v4.1-flash
glm-5.3
deepseek-v4-pro-0813
glm-5.3-flash
qwen3.8-27b
qwen3.8-2.4t-a95b
kimi-k3
glm-5.2
kimi-k2.7-code
minimax-m3
gemma-4-31b-it
gpt-oss-120b
nemotron-3-ultra-550b-a55b
qwen3.6-35b-a3b
Throughput and TTFT are live medians, measured continuously.
Top up, spend against it, auto-refill when low.
Just plug in and start building. Stop anytime.
High-volume discounts, dedicated support and SLAs at scale. Talk to us ↗
One published rate per model, billed per token, the same rate no matter which provider or node serves the request. Keln takes no platform fee and adds no per-provider markup, so your cost is predictable even as routing changes underneath you.
No. Price is fixed per model, so Keln is free to send every request to the fastest verified capacity. Faster nodes cost you the same as slow ones.
Yes. Every model lists an input and an output rate per 1M tokens. Reasoning tokens are billed as output tokens, and cached input is billed at a reduced rate.
Prepaid credits. Top up a balance, spend against it, and set auto-refill so you never hit zero. Spend is visible live per project and per key. Keln auto-generates invoices for business.
You are billed once, for the tokens you receive. Failed attempts and Keln-initiated failovers are on us. Recovery is part of the service.
Contact us ↗ to get more information on high-volume discounts.