Kite Launches Marathon Adaptive Reasoning Infrastructure for Long-Running AI Agents
2026-07-22 10:22

Woofun AI reports that Kite has launched Marathon, an adaptive reasoning infrastructure designed for long-running AI agents. The platform allows users to select completion windows ranging from immediate to flexible timing, with pricing decreasing as wait times increase; the deepest tier offers approximately 65% savings compared to real-time rates. The service API is compatible with OpenAI and includes a one-line installation plugin for Claude Code and Codex.

Initial support covers five flagship open-weight models: Kimi K3, GLM 5.2, DeepSeek V4 Pro, Qwen3.6-35B-A3B, and Nemotron 3 Ultra, all adhering to the same pricing structure. This launch follows Kite’s $33 million financing round led by PayPal Ventures and General Catalyst, marking a strategic expansion from agent identity and payment networks into underlying computational infrastructure for the Agent Economy.

Disclaimer: Views are the author's own and do not represent the platform. Do not reproduce without permission. Content is for reference only, not investment advice. Trade at your own risk.
Tags:
Kite
Marathon
OpenAI
Claude Code
Codex
Kimi K3
GLM 5.2
DeepSeek V4 Pro
Qwen3.6-35B-A3B
Nemotron 3 Ultra
PayPal Ventures
General Catalyst
Share:
back