# What's new in Veqen 1.5

## Use Veqen anywhere

- **The API covers nearly everything the dashboard does.** With a full-access key: spikes, savings, runs, forecasts and the bill check; the AI analyst; and budgets, credits, bills, your prices, month-end, connections, saved views, alerts, members, send-only keys, the audit log and the limit on your Veqen bill. [API reference](/docs/api)
- **More MCP tools:** budgets, savings, one run or spike, credits, bills, your prices, month-end, alerts and members, under the signed-in person's role. [MCP server](/docs/mcp)
- **Safer full-access keys:** a key acts as an admin only while the person who created it is still an owner or admin, can't make other full-access keys, and its changes are marked in the audit log. [API keys](/docs/api-keys#what-a-full-access-key-can-do)

## A clearer dashboard

- **Pages named for what you do there:** Overview, Breakdown, Customers, AI models and Agent runs; Spikes, Savings and Forecast; Cost planner and Budgets; Bill check, Provider credits and Month-end. In Settings, Custom prices is now **Your prices** and Connected apps is **AI assistants**.
- **(i) badges** explain terms where they appear, and **tips** sit next to the decisions they help with. Every page links to its guide.
- **Get started**, in the sidebar: six steps, ticked off by what your organization actually does. [Quickstart](/docs/quickstart#the-get-started-checklist)
- **The AI analyst answers in a fixed shape:** the answer, why, and what to do, in dollars and cents, with suggested questions, readable charts, and the numbers behind it one click away. It can now look up savings, one run and one spike. [AI analyst](/docs/ai-analyst)
- **Describe your product** in the cost planner: examples to start from, a summary of what was drafted and assumed, and Undo. [Simulator and cost planner](/docs/simulator)
- **Cheaper model suggestions** only name current, stable models, never previews or dated snapshots.

## Know what moved, and why

- **Anomalies by dimension.** A spike on all spend, or on one customer, user, feature, project or model, is caught the hour it happens and alerted once; its page shows what drove it. [Insights](/docs/insights)
- **The weekly digest explains itself:** new or stopped features and customers, ones whose spend tripled, a shift in model mix, a cache hit rate that fell, and models with no price.
- **Savings:** prompt caching, the Batch API, the same open model on a cheaper host and cheaper models from the same provider, each with a monthly estimate and a link to the cost planner.
- **Forecasts** for the month and quarter with an 80% range, and budget pace warnings before a monthly budget runs out.
- **Runs:** tag `run` and see what each agent run or conversation cost, how run costs spread, and the costliest ones call by call. [Runs](/docs/runs)
- **The AI analyst draws charts**, and every anomaly has an Explain box. [AI analyst](/docs/ai-analyst)

## Every bill, reconciled

- **Cost imports from AWS Bedrock, Google Cloud (Vertex AI and Gemini), Azure (OpenAI and AI Foundry) and OpenRouter**, beside OpenAI and Anthropic; AWS, Google Cloud and Azure can each be split by a tag or label. Anthropic's Priority Tier is priced from its usage report. [Reconciliation](/docs/reconciliation)
- **Restatements** are tracked, and flagged when they touch a month you've closed.
- **Billing exports** from any cloud or GPU host, by month and in any currency, and **Use as price** to price self-hosted models from their bill.
- **Rate cards:** upload a contract's prices as a CSV. [Custom prices](/docs/custom-prices)
- **Calls count toward the host that bills them:** Azure OpenAI, Claude on Bedrock and Vertex, and models through OpenRouter.

## Finance

- **Departments, classes and legal entities** from your tags, in the close and the QuickBooks and Xero exports. [Monthly close](/docs/close)
- **True-ups** instead of reversing accruals, if you prefer: post just what changed after a month was approved.
- **Books in your currency** (GBP, EUR, CAD, JPY and the rest of the ECB's list) at the month-end rate.
- **Margin counts credit notes and refunds,** and revenue in any currency the ECB publishes. [Margin](/docs/margin)

## Alerts and reports

- **Alerts are queued and retried** for about a day, with a log of every alert sent. [Alerts and reports](/docs/alerts)
- **Saved views** in Breakdown, emailed weekly or monthly.

## Security and enterprise

- **API keys** that expire, work only from listed addresses and have their own rate cap, with rotation and reminders. [API keys](/docs/api-keys)
- **SAML or OIDC single sign-on and SCIM directory sync** on Enterprise, with SSO required for your verified domains. [SSO](/docs/sso)
- A measured **[status page](/status)**, a **[trust page](/trust)**, an **[uptime SLA](/legal/sla)** and a **[DPA](/legal/dpa)**.

## Public pages

- **[Integrations](/integrations)**: a page per framework and host with working code.
- **[AI price index](/price-index)**: a monthly report of price moves.
- **Shared simulations** open with a preview card when you post the link.

## SDKs

Python 1.3.0 and JavaScript 1.1.0:

- the `run` tag; the OpenAI Agents SDK tags each agent run;
- a LlamaIndex event handler (`veqen[llamaindex]`);
- Azure OpenAI calls are reported as `azure-ai`, Claude through `AnthropicBedrock` and `AnthropicVertex` as `bedrock` and `google`, and calls through OpenRouter as `openrouter`, so they reconcile with those bills. If you filter on the provider `openai` for Azure traffic, switch to `azure-ai`;
- Bedrock cross-region profile ids (`us.`, `eu.`, `global.`) are priced at Bedrock's own rates.