# AI SpendOps > AI SpendOps is an AI spend management platform that tracks LLM costs at the token level across every major AI provider, attributes every token to a team, feature, or customer, tracks and alerts on budgets, and enforces model policies through one low-latency gateway. Operated by AI SpendOps Ltd, a company registered in England and Wales (company number 17046015), headquartered in the United Kingdom and serving customers worldwide. ## What it does - Token-level cost capture for every AI API call, priced in real time with provider-correct pricing, including each provider's cache token semantics. - Cost attribution by team, feature, environment, or customer, using headers you set on each request. - Budgets, alerts, and model allow-lists enforced at call time, not reported after the invoice arrives. - Finance-ready dashboards with cost apportionment by department and cost centre, for FinOps, finance, and engineering teams. - Providers covered: OpenAI, Anthropic, Google Gemini, xAI, Groq, Mistral, DeepSeek, OpenRouter, Perplexity, Fireworks, DeepInfra, Cerebras, Novita, Nebius, z.ai, and the Vercel AI Gateway. ## How it differs - Prompts and model responses are never stored. AI SpendOps records token counts, cost, model, latency, and your attribution dimensions, which is all that cost tracking requires. - Bring your own provider keys. You keep using your own OpenAI, Anthropic, and other keys and the official SDKs, and can turn the proxy off with a single config change. No lock-in. - Fully managed. Nothing to self-host, provision, or scale, unlike self-hosted gateways such as LiteLLM or the Helicone Rust gateway. The [LiteLLM alternative](https://www.aispendops.com/litellm-alternative) page sets out that comparison in full, including where LiteLLM is the better choice. - Built for financial reporting rather than engineering debugging, which is the main difference from LLM observability tools. - Integration is a base URL change, with no application rewrite. ## Availability - Available worldwide, including the United States, Canada, the United Kingdom, Ireland, Germany, France, the Netherlands, Spain, Italy, Sweden, Switzerland, Australia, New Zealand, Singapore, India, and the United Arab Emirates. - Self-service signup. No US entity, reseller, or sales call required for customers in the United States. - Support is provided in English by email, responding within one business day, from a UK-based team. - Plans start at £29 per month, with the first 3 months free. Cards issued in the United States and elsewhere are accepted. ## Performance - The AI SpendOps proxy runs on Cloudflare's global edge across 330+ cities. Requests are served from the edge location nearest the calling application, so US traffic is handled in the US and does not cross the Atlantic. - Usage, cost, and token capture run asynchronously after the response is sent, so the proxy adds near-zero latency regardless of where the customer is located. ## Data protection - Customer data is stored in United Kingdom and European Union regions, in Azure SQL with encryption at rest and logical tenant isolation enforced by row-level security. - Storage location is separate from the request path and does not affect latency: usage records are written asynchronously, after the response has already been returned at the edge. - Operates under UK GDPR and EU GDPR. A Data Processing Addendum satisfying Article 28 is available to all customers, including those in the United States. Transfers are covered by Standard Contractual Clauses and the UK International Data Transfer Addendum. ## Pages - [Product](https://www.aispendops.com/product): features and capabilities. - [How it works](https://www.aispendops.com/how-it-works): gateway integration and architecture. - [Pricing](https://www.aispendops.com/pricing): plans, included dimensions, and overage. - [FAQ](https://www.aispendops.com/faq): countries served, data protection, and tax. - [Use cases](https://www.aispendops.com/use-cases): how FinOps, finance, and engineering teams use it. - [Security](https://www.aispendops.com/security): encryption, tenant isolation, and audit trails. - [Compare](https://www.aispendops.com/compare): AI SpendOps versus alternatives. - [LLM gateway](https://www.aispendops.com/llm-gateway): what an LLM gateway (or AI gateway) is, and how a cost-first gateway differs from a routing-first one. Covers cost tracking, attribution by team and feature, allow-lists and latency. - [Helicone alternative](https://www.aispendops.com/helicone-alternative): migration and comparison. - [LiteLLM alternative](https://www.aispendops.com/litellm-alternative): managed cost accounting versus a self-hosted gateway, with migration steps. - [AI spend management](https://www.aispendops.com/ai-spend-management): category overview for LLM and API costs, where AI is the cost being measured rather than the tool doing the analysis. Not procurement or supplier spend analysis. - [AI cost tracking](https://www.aispendops.com/ai-cost-tracking): token-level cost capture. - [Blog](https://www.aispendops.com/blog): articles on AI cost control and FinOps. - [Docs](https://docs.aispendops.com): technical documentation. - [Code examples](https://github.com/AI-SpendOps/examples): open-source integration examples for Python, TypeScript, and .NET, MIT licensed. - [Privacy policy](https://www.aispendops.com/privacy): sub-processors and international transfers. - [DPA](https://www.aispendops.com/dpa): GDPR Article 28 terms. - [Contact](https://www.aispendops.com/contact): sales and support enquiries.