What's new in Revenium 2.15: broader coverage, sharper insights

The last three releases (2.13, 2.14, and 2.15) brought a lot of momentum to Revenium. New providers, smarter analytics, and a handful of reliability and governance upgrades that customers had been asking for. Here's a roundup of what landed.

Five highlights

1. Cost analytics by AI job

The Traces page now has a Job view that aggregates cost, anomalies, and trends by agentic job instead of by individual trace. You can see average and P95 cost per job, your most expensive jobs, and cost trends by job type. You can also open any job and view its transactions grouped by trace.

If you're running agentic workloads, this is the view that finally tells you which jobs are quietly running up your bill, not just which individual calls.

2. AI Insights catches new classes of waste with dollar estimates

AI Insights now detects two new patterns that quietly drain budgets:

• max_tokens cutoffs — responses being truncated mid-thought while you pay for incomplete answers
• Dormant subscribers and agents — users or agents whose AI spend has gone largely inactive but still has provisioned access

Every finding comes with a dollar-quantified savings estimate and example transactions so you can verify it before you act. And in 2.15, insight cards became click-through: tap an insight and land directly in the filtered transactions behind it.

AI Insights is also now available through the MCP server, so AI agents talking to Revenium can run and read recommendations as a first-class tool action.

3. Azure Cost Management integration

You can now connect your Azure subscription to Revenium using a Service Principal and pull in daily cost data. Spend is attributed to the specific model that drove it, matching the per model breakdown you already get for OpenAI and other providers. Credentials support pasting JSON, and Azure billing errors come with clear and actionable guidance.

4. GitHub Copilot support

Track your organization's Copilot usage in Revenium alongside your other AI coding assistants. Connect with the Revenium CLI (revenium-copilot) and see adoption and usage on the new Copilot dashboard. In 2.15, we also updated the CLI to track Copilot through GitHub's current Metrics API, so reporting stays reliable.

5. Cursor cost tracking is now zero-setup

Paste your Cursor team API key into provider settings and Revenium pulls usage every five minutes directly from Cursor's API. No middleware, no agents, no config, just connect and go.

Also in these releases

Governance and security

• Role-restricted API keys — creating a write-scope API key is now restricted by role, so you can let people issue read-only keys without granting write access (2.15)
• HMAC signing for outbound webhooks with safe rotation and a one-click test-event button (2.13)
• Webhook verification helpers in every SDK — verify HMAC-signed webhooks from Revenium with a single SDK call in Python, Node, or Go (2.14)

Metering reliability

• Idempotency-Key support is now on by default on metering endpoints, with the Node, Python, Go, and MCP SDKs generating the key automatically. Retried submissions within 24 hours are deduplicated server-side (2.13 + 2.14)
• X-RateLimit-* response headers and a documented /integrations/rate-limits reference, so customers can program against per-tenant request budgets (2.13)

SDKs and middleware

• Python SDK gained a built-in agentic outcomes API, cost-controls + circuit breaker integration, and a renamed BudgetExceededError for clearer error handling (2.13)
• Python SDK fix for AWS Bedrock: cache token counts on the Bedrock Anthropic path are now included (2.14)
• Node middleware now uses separate metering and enforcement circuit breakers, so an enforcement issue can't stop metering (2.13)

Pricing modes and team controls

• AI Assistant Pricing Mode — a per-team setting (Management > Teams > Team AI Settings) that controls whether AI assistant usage counts as real API spend or stays as a subscription/seat-based estimate on AI Assistant dashboards, so your cost reports reflect what you actually pay (2.14)
• Coding-assistant pricing mode in Team settings — choose whether coding-assistant traffic is paid for using API rates or seat-based pricing (2.13)

For more details on improvements we’ve made in these releases, including minor updates and bug fixes, check out our docs site to see the changelog.

Try it out

All of the above is live for every Revenium customer. If you're already on Revenium, the new Job view in Traces and the updated AI Insights are probably the fastest place to see immediate value. They surface opportunities you did not have to go looking for.

Want to see these capabilities in action? Join our founders on June 25 for a live walkthrough of the new Job view, AI Insights, and the cost patterns quietly draining AI budgets. Save your seat.