Pricing

Sign up and your organization gets three days of Pro features straight away, with no card or payment details. Confirm your email address to lift the trial caps: until then the trial is limited to 1,000 gateway requests and 10,000 traced spans. To continue after the trial, use Request upgrade in the console or email us, and we activate your plan.

If no plan has been activated, the gateway and telemetry ingestion stop accepting traffic and answer 402 subscription_inactive. You can still sign in and read everything in the console. Your organization, configuration and data stay in place (telemetry keeps following your plan’s retention period), and traffic resumes as soon as a plan is active.

We don’t take card payments online yet. Use Request upgrade in the console (Settings → Plan & billing) or email info@xeoo.io with the plan and billing interval you want. We confirm the price, arrange payment with you and activate the plan on your organization.

No. You connect your own providers, from Anthropic and OpenAI to DeepSeek, Qwen or any OpenAI‑compatible server, and each provider bills you directly for model usage. Worksi never resells or marks up tokens: our plans cover the control plane only.

Each model call your application sends through the gateway that is forwarded to your provider counts once: a chat completion, a Responses API call, an Anthropic Messages call or an embeddings call. A call that a fallback model answers still counts once. Requests the gateway blocks, whether by a policy, a quota, a paused agent or a stopped run, are not counted.

A span is one recorded unit of work: a model call, a tool call, an agent step, a retrieval, an approval wait, or the start and end of a run. A gateway call records one span; a fully traced workflow run typically records 10 to 20. Spans from the SDKs, OpenTelemetry and MCP all count.

When an incident opens, Worksi runs deterministic checks and can then ask a model to reason over the evidence, through the platform’s OpenAI key or a credential of yours. Diagnoses made with the platform key count toward your plan’s monthly allowance. Use your own OpenAI or OpenAI‑compatible credential and they are unlimited. Deterministic diagnosis is always included.

Traces and their previews are kept for your plan’s retention period: 14 days on Starter, 30 days on Pro and the trial, 90 days on Business, and 365 days or more on Enterprise. If your organization sets a shorter period, that applies instead. The audit log is kept permanently.

We notify your organization’s owners at 80% and 100% of each quota. At 100% of gateway requests or spans, new traffic is refused with 429 plan_quota_exceeded until the next calendar month (UTC) or until a larger plan is activated.

Yes, on every plan. End‑user plans let you pass each end user and their plan with two headers, and an end‑user quota policy limits requests, tokens, cost and models per user.