Outrun AI
A prepaid AI API giving one key and one USD balance access to GPT, Claude, Gemini, Grok, DeepSeek and other text, image and video models.
At a Glance
About Outrun AI
Outrun AI is a prepaid AI API at outrun.to, built and operated by Outrun LLC, a Wyoming-registered US company. One key and one USD balance reach models from OpenAI, Anthropic, Google, xAI, DeepSeek, Zhipu, Moonshot, MiniMax and Alibaba, plus uncensored models, through an OpenAI-compatible endpoint. The site lists 97 models (121 model IDs including alternative routes) and says the service has been live since October 2026.
What It Is
Outrun is an API gateway and billing layer for AI models. Developers and power users sign in, copy a personal API key, add credit, and point their tools at the base URL https://api.outrun.to/v1. They then choose a model ID. The service also exposes Anthropic-style Messages and Gemini-native endpoints. Text, image, audio and video models are available from the API, a web Playground or a Telegram bot.
Model Catalog and Uncensored Options
The catalog covers GPT, Claude, Gemini, Grok, DeepSeek, GLM, Kimi, MiniMax and Qwen families, plus image and video models such as Seedance. Of the 97 models, 39 are described as uncensored or abliterated, meaning tuned or modified to refuse far less. Outrun notes that an app's own system instructions can still affect answers, and that users remain responsible for lawful use.
Coding Tool Setup
Documentation provides setup guides for Codex (CLI and desktop app), Claude Code, Gemini CLI, OpenCode, Grok Build CLI, Cline, Roo Code, LiteLLM, SillyTavern, Janitor AI, Zo Computer and mobile apps. Each guide lists the connection values and a test step. Outrun also publishes a guide for continuing to code on Outrun after hitting a provider usage limit.
Pricing Transparency and Privacy
Outrun states that it publishes per-model prices, including a JSON feed, and compares them with official list prices. These savings figures are vendor claims. It reports running baseline model audits every 14 days with promptfoo and a Meow LLM Detector check on some routes, and says the model ID you choose is the model that answers. It says it does not store prompts, images or outputs unless users choose to save them.
Community Discussions
Be the first to start a conversation about Outrun AI
Share your experience with Outrun AI, ask questions, or help others learn from your insights.
Pricing
Prepaid pay-as-you-go
Prepaid USD credit billed per token, image or second of output; no subscription or monthly fee. Credit can be added by card or crypto from $5.
- GPT-6.1 Sol: $0.12 input / $0.59 output per 1M tokens (up to 272k tokens), cached input $0.01
- GPT-6 Astra: $0.59 input / $2.94 output per 1M tokens (up to 272k tokens), cached input $0.06
- Claude Sonnet 5.5: $0.49 input / $2.45 output per 1M tokens, cached input $0.03
- Gemini 3.8 Flash: $0.32 input / $1.58 output per 1M tokens, cached input $0.04
- Grok 4.7: $0.09 input / $0.26 output per 1M tokens (up to 200k tokens), cached input $0.03
- Lowest text price: Claude Haiku 5.5 (2) at $0.02 input / $0.10 output per 1M tokens
- Images charged per image, from $0.02
- Video charged per second of output, from $0.03
- Playground web search: 10 free searches per account each day, then $0.0015 per successful search
- Add credit by card from $5 or crypto from $5, or redeem a prepaid code
Capabilities
Key Features
- One API key and one USD balance for all models
- OpenAI-compatible endpoint plus Anthropic Messages and Gemini-native endpoints
- 97 models including GPT, Claude, Gemini, Grok, DeepSeek, GLM, Kimi, MiniMax and Qwen
- 39 uncensored and abliterated models
- Image, video and speech generation billed per image or per second
- Web Playground and Telegram bot
- Published per-model pricing and public pricing JSON feed
- Periodic model audit results published
- Pay by card, crypto or prepaid code
- Setup guides for coding agents and chat apps
