Flatkey AI

Flatkey AI is a hosted AI API gateway. Its official gateway page describes one OpenAI-compatible key and base URL for routing requests to models including GPT, Gemini, Claude, DeepSeek, Kimi, and Seedance. The platform combines model access with automatic failover, usage analytics, a shared balance, and a single invoice. Its homepage also presents 1,000+ pay-per-call data APIs and MCP tools, while the CLI and governance pages target agent and production workflows.

✨ Key Features

• Unified model endpoint — Keep an OpenAI-compatible client and switch models by changing the model identifier instead of maintaining separate provider integrations.
• Multi-provider access — Route supported text, image, video, and other model requests through one key and one balance.
• Automatic failover — Health-check upstream channels and retry or reroute failed requests without client-side changes.
• Usage analytics — Inspect per-key and per-model usage, live spend, request logs, and plan limits.
• Unified billing — Consolidate provider usage into one balance and invoice for simpler procurement and accounting.
• Model pricing directory — Compare supported model endpoints and published usage rates before choosing a route.
• Tools and MCP access — Call 1,000+ pay-per-call data APIs and MCP tools through Flatkey Credits where available.
• Agent and CLI workflow — Use the published SKILL.md and CLI-oriented setup to connect coding agents and other agent tools.
• Team governance — Enterprise materials describe sub-key budgets, model allowlists, alerts, ledger APIs, and procurement support.
• Reliability and data controls — The public pages advertise a signed SLA, zero retention of request content, and audited-infrastructure controls; verify scope by plan and current terms.

💳 Pricing

The official pricing page currently lists monthly self-serve plans and an enterprise option. It also says new users start with a US$1 free credit, and top-ups add balance after plan usage is exhausted.

• Starter — Promotional US$10/month (the page also shows a crossed-out US$45 reference); for individuals and light daily use. Includes all models, tools credits, and pay-per-run usage.
• Pro — Promotional US$30/month (with a crossed-out US$90 reference); for daily development and higher-frequency requests. Includes all models, tools credits, and pay-per-run usage.
• Max — Promotional US$100/month (with a crossed-out US$300 reference); for teams and heavy workloads. Includes all models, tools credits, and pay-per-run usage.
• Enterprise — Custom pricing for committed volume, with committed-volume discounts, custom routing, invoicing, and procurement support.

Top-ups are available separately: the pricing page advertises US$10, US$30, and US$100 top-up examples with promotional bonus credits. The public page does not provide a single fixed allowance for every model; actual cost depends on model and request usage. Confirm promotional validity, quotas, taxes, and model rates before purchase.

⚖️ Alternatives Compared

🎯 Use Cases

• AI product teams — Add multiple model providers behind one API surface and reduce integration changes when testing models.
• Backend engineers — Keep an OpenAI-compatible SDK while centralizing keys, routing, and spend visibility.
• Agent builders — Give coding agents or automation tools access to supported models and pay-per-call tools through one balance.
• Reliability engineers — Use health checks, retries, and failover to reduce dependence on a single upstream channel.
• Finance and procurement teams — Reconcile model usage through one balance, request ledger, and invoice.
• Platform teams — Apply sub-key budgets, model allowlists, and usage monitoring for teams or environments where available.
• Media developers — Route supported image and video generation requests alongside text-model calls.
• Enterprise buyers — Discuss committed volume, custom routing, invoicing, and procurement requirements with sales.