OfoxAI

What it is

OfoxAI (ofox.ai) is a multi-model API gateway: one OpenAI-compatible key and base URL for a large catalog of text, image, and video models (they claim 100+). Verified 2026-09-22 from the live site: swap the endpoint, keep your existing SDK/tooling (Claude Code, Codex, and others are named). Catalog includes open-weight lines (Qwen, DeepSeek, GLM, Kimi) and frontier brands, plus image/video generation endpoints with listed per-image and per-second prices. They advertise multi-upstream failover, ~300ms edge latency claims, and “routing, not storage” as a privacy posture — treat uptime/latency numbers as vendor SLAs to verify.

Who it’s for

Builders who want to A/B models or add regional open-weight options without rewriting clients. Useful when you’re prototyping agents and don’t want five vendor keys yet. Less ideal if you need a signed BAA with a single frontier lab, or if procurement requires direct vendor contracts only.

Why we listed it

Appeared in Shipstry’s Sep 7–13, 2026 ship week; live URL verified 2026-09-22. Early multi-model gateway with a clear maker workflow (one base URL). Matches how indie agents actually get wired.

Caveats

Gateway = another party in the data path. Read their retention/routing docs before sending customer content. List prices and model IDs churn weekly in this category — copy rates into your own sheet the day you budget. We have not independently verified the 99.9% uptime claim or every model card on the catalog. Confirm ToS for commercial use of generated media.