Guides
Guides for choosing and governing AI models
Start from the question you actually have: which model, at what cost, under which controls. Every guide uses current prices and cites vendors.
02 / Govern
Run it as a team
Step-by-step setups from the docs: keys per project, budgets that stop spend, and where the money went.
03 / Reading
Further reading
Claude Code cost for teams: what a developer costs per day and 10 controls (2026)Claude Code cost averages about $13 per developer per active day. A 25-developer budget worked out, plus 10 controls that keep agent spend inside a ceiling.AI API spending limits compared: OpenAI, Anthropic, OpenRouter, Vercel and ATP Token (2026)AI API spending limits compared: how OpenAI usage limits, Claude workspace caps, OpenRouter, Vercel and ATP Token behave at the limit, with a worked budget.Model catalog vs access control: GET /v1/models lists the menu, the project allowlist decides (2026)GET /v1/models lists what an LLM platform offers; a model allowlist decides what a key may call. Request and response examples, the 403, and allowlist design.API key management for LLM APIs: one project, one key, and a 30-day migration runbook (2026)API key management for LLM APIs: what breaks with a shared key, how to issue one key per project, where each key lives, and a 30-day migration runbook.LLM gateway comparison 2026: OpenRouter, Vercel AI Gateway, LiteLLM, Portkey, Helicone and ATP TokenWhat an LLM gateway does, and how six gateways compare on hosting, pricing model, spend limits, model allowlists and API formats as of October 2026.OpenRouter alternatives for teams: pricing, budgets and model controls compared (2026)OpenRouter alternatives compared: what OpenRouter charges, how its budgets and allowlists work, and when LiteLLM, Vercel AI Gateway or ATP Token fit better.
Get started
Try both on one key — start with USD 5.
Issue a project key, allowlist the models you want to test, and compare them on your own traffic.