app.nz | AI Agent Cloud for Coding Agents, Deploy, Models
repos · tested PRs · deploys · hosted models<br>The AI agent cloud for shipping software.<br>Start with a repo task. app.nz runs the coding agent, proves the change with tests and visual checks, opens the PR, then hosts the app, worker, API, or Cog endpoint from the same account.<br>Start building Explore platform<br>Coding agentsRepos & PRs15+ model providersApp & GPU hostingWeb & paper search
start fast<br>Create Apps<br>Spin up a project with repos, models, and storage. Ship a full-stack AI app, no infra wiring.
autonomous<br>Coding Agents<br>Give an agent a repo task. It creates a branch, edits code, runs tests, and opens a PR for review.
git hosting<br>Repos & PRs<br>Hosted repos with GitHub sync, branch previews, review queues, and agents that repair failing CI.
chat & search<br>Chat & Research<br>Ask questions, attach files, run deep research, and keep every citation in one workspace.
agent tools<br>Search the Web<br>Fresh web results, summaries, and citations for your agents and apps, through one API key.
research<br>papers.app.nz<br>Search 200M+ papers, methods, citations, and code from one account. Built for R&D and agents.
model routing<br>Models<br>One OpenAI-compatible API for frontier, fast, cheap, vision, image, and code models. Auto-route or pin a provider.
deploy<br>Hosting<br>Deploy static sites, containers, APIs, workers, and model endpoints with HTTPS preview URLs on app.nz.
GPU inference<br>Cog Studio<br>Deploy Cog containers as scale-to-zero GPU endpoints with generated forms, prediction logs, and clear billing.
container CI<br>Build Studio<br>Build Docker images on app.nz workers, push to your private registry, then run them in Cog Studio or deploys.
task queues<br>Queues<br>Push JSON jobs to a queue and let agents or workers process them with retries, dead-letter, and spend caps.
CLI for the agent cloud<br>Start agents, deploy, and route models from your terminal.<br>The app CLI runs agents, builds images, ships deploys, rotates keys, and calls the model gateway. Same workflows as the web app.<br>$curl -fsSL https://app.nz/cli.sh | shView script
macOS, Linux, and Windows builds
Create a repo<br>$ app repo create acme/ai-dashboard --public<br>Initialized repo acme/ai-dashboard<br>Added README, .gitignore, license<br>Pushed to origin/main<br>Repository created
Run an agent<br>$ app agent run "add usage analytics" --repo acme/ai-dashboard<br>Planning task<br>Creating branch feat/usage-analytics<br>Writing code and tests<br>Opening PR #42
Deploy<br>$ app deploy --app ai-dashboard --region sjc<br>Building<br>Running checks<br>Deploying to production<br>https://ai-dashboard.app.nz
Cloud coding agents<br>Autonomous agents that code in your repos and ship through CI.<br>Hand off a task in plain English. Agents branch, build, test, open pull requests, and fix red CI. You see the plan, logs, changed files, costs, and review link before anything merges.
Plan and implement<br>Each agent gets a branch, sandboxed terminal, full repo context, and tools to iterate until the change is ready.
Prove it before merge<br>Tests, lint, security checks, and reviewer passes all run before a PR lands.
VisualBench proof<br>Agents screenshot the UI, compare the result, and fix their own work until the interface matches what you asked for.
Repair CI automatically<br>When checks fail, agents read logs, patch the regression, and rerun jobs without you copying errors by hand.
Research for builders<br>Chat, live web search, and deep research in one place.<br>Ask a question, attach files, pull live web results, run multi-step research, or search papers, then feed the answer into your next agent task. Every source stays in your account.
Research Chat<br>web + papers + files + model routing
one workspace
Chat<br>Fast assistant conversations with uploads, images, model selection, and conversation history.
Web Search<br>Ground answers with live web results, highlights, citations, freshness controls, and provider choice.
Deep Research<br>Multi-step research plans that search, fetch, summarize, cross-check, and produce source-backed reports.
Papers<br>Search papers, methods, datasets, GitHub code, citations, and recent work through papers.app.nz.
POST /v1/search<br>{"provider":"exa","type":"deep","query":"compare hosted model pricing with citations"}<br>POST /api/research/start<br>{"message":"research R2-backed model hosting options","web_search":true,"papers":true}
Models<br>One gateway. Route providers or host your own weights.
Point any OpenAI-compatible client at app.nz and let app/auto pick the model, or pin a provider. Host custom weights, LoRAs, and datasets on R2 storage to run inference on your own machines.
frontier<br>app/auto<br>general chat, planning, mixed workloads<br>auto-routed
agentic<br>app/auto-code<br>repo edits, PR repair, migrations<br>best-fit
latency<br>app/auto-fast<br>realtime UX, chat, tools<br>low-latency
research<br>app/search-deep<br>web search, papers, citations, extraction<br>$0.001+ per search
media<br>app/auto-image<br>images, edits, art workflows<br>per output
hosted<br>app/model-hosted<br>custom R2-backed model...