AI Decision System
Replayable, offline-verifiable records for AI-assisted go/no-go decisions: experiment promotions, typed model answers and human-approved actions.
翻译准备中
个人 GitHub 账号下公开仓库的所有者可以登录并验证所有权,自行管理介绍,无需管理员批准。
登录并生成验证文件,在 30 分钟内将其提交到此仓库的默认分支。其他分支或分叉仓库中的文件无效。然后在下方检查文件,即可无需管理员批准直接管理项目。请保留文件以维持管理权限。
如需登录,将前往登录页面。登录后请回到此处,再次点击所选的验证按钮。
在项目文件夹中运行 /nomorevibe verify,即可验证所有权并自行管理介绍。
一分钟完成验证Explore models form Vercel AI Gateway. Filter and compare by provider, context window, and pricing.
翻译准备中
取自仓库 README
正在加载完整 README。
查看完整 README ↗https://github.com/user-attachments/assets/bfc21467-2d39-47fd-8667-cc359269175d
A model explorer website for AI gateways. Auto-fetches the latest models from Vercel AI Gateway and lets you browse, compare, and filter them.
Live site: ai-registry-website.vercel.app
This monorepo contains:
apps/website: a Next.js model explorer that fetches models at runtime from supported AI gateways (currently Vercel AI Gateway), with caching and a build-time fallback snapshotPrerequisites:
packageManager: [email protected])Install dependencies at the root:
pnpm install
Common scripts (powered by Turbo):
pnpm devpnpm buildpnpm check-typesRun the website:
pnpm --filter website dev
The website fetches models on the server using unstable_cache (1-hour TTL, tag-based revalidation). The active gateway is resolved via a registry abstraction in apps/website/lib/ai/gateways/, currently providing Vercel AI Gateway. Adding more gateways is just a matter of implementing the GatewayProvider interface and registering it.
If the live fetch fails (e.g. during build with no network), a generated snapshot at apps/website/lib/ai/models.generated.ts is used as a fallback. Refresh it with:
pnpm fetch:models
The Vercel gateway requires an AI_GATEWAY_API_KEY (or VERCEL_OIDC_TOKEN) at runtime/build time when fetching live data.
MIT. See LICENSE for details.