Open inference infrastructure
for autonomous agents.

A single OpenAI-compatible endpoint for DeepSeek, Qwen, Kimi, Minimax, GLM, and more.

models →
deepseek/deepseek-r1deepseek/deepseek-r1-0528deepseek/deepseek-v3.2deepseek/deepseek-v3.2-specialedeepseek/deepseek-chat-v3.1deepseek/deepseek-chat-v3-0324qwen/qwen3-235b-a22b-thinking-2507qwen/qwen3-coder-nextqwen/qwen3.5-397b-a17bmoonshotai/kimi-k2-0905moonshotai/kimi-k2.5minimax/minimax-m2.1minimax/minimax-m2.5z-ai/glm-4.7z-ai/glm-5openai/gpt-oss-120bdeepseek/deepseek-r1deepseek/deepseek-r1-0528deepseek/deepseek-v3.2deepseek/deepseek-v3.2-specialedeepseek/deepseek-chat-v3.1deepseek/deepseek-chat-v3-0324qwen/qwen3-235b-a22b-thinking-2507qwen/qwen3-coder-nextqwen/qwen3.5-397b-a17bmoonshotai/kimi-k2-0905moonshotai/kimi-k2.5minimax/minimax-m2.1minimax/minimax-m2.5z-ai/glm-4.7z-ai/glm-5openai/gpt-oss-120b
deepseek-r1deepseek-r1-0528deepseek-v3.2qwen3-235b-thinkingqwen3-coder-nextkimi-k2kimi-k2.5minimax-m2.5glm-4.7glm-5gpt-oss-120bminimax-m2.1
01

Fresh open models, fast

New open source releases land on Rungate immediately after they are published. Whatever ships next will be too.

Browse models →
02

High uptime for long-horizon runs

Autonomous agents cannot afford mid-task failures. Rungate keeps throughput high and latency consistent so your agents complete what they start.

Learn more →
ThroughputLatencythroughputlatency
CLIENTOpenAI SDK · OpenClawany OpenAI-compatible clientrungateAUTHx402 · API Keypay per call or authenticate
03

Works with what you use

Connect via OpenClaw skills or any OpenAI-compatible client. Pay per request with x402, or authenticate with an API key.

View docs →
04

Security for autonomous runs

Detect and block prompt injection before it reaches your agent. Evaluate model skills and set capability boundaries for unsupervised runs.

Read the Thesis →
promptinjectionmaliciousskillsrungatefilterdetected

Questions? info@rungate.ai