August 2026 release is live Read More
Breadcrumbs

LiteLLM Onboarding

LiteLLM is an open-source gateway that gives you one consistent, OpenAI-compatible interface for calling 100+ LLMs — OpenAI, Anthropic, Azure, Bedrock, and more — without rewriting code per provider. Most teams run it as a self-hosted proxy to centralize auth, routing, and spend tracking across all their LLM usage.

Why this matters: when teams call multiple LLM providers directly, tracking spend means checking each provider's own billing console separately — and that's assuming you even know every provider in use, since anyone with an API key can add a new one. LiteLLM already centralizes the routing; connecting it here centralizes the cost visibility too, in one place, across every model behind it, regardless of which provider actually served the request.

This shows up differently depending on your role:

  • FinOps managers use this to get one combined view of LLM spend across every provider behind the gateway, instead of stitching together numbers from each vendor's console.

  • Platform/engineering teams run the proxy and do the actual connection — editing the config, setting the environment variables, and keeping it on a supported version.

  • Finance leaders use this for chargeback when the gateway serves multiple teams or products, without needing per-provider breakdowns done manually.

Connect it here so that cost data flows into Mavvrik automatically, across every model routed through it — LiteLLM exports it as FOCUS 1.2 formatted cost reports, the FinOps Foundation's open cost-and-usage standard. This integration requires LiteLLM v1.91.0 and above.

Before you start: you'll need shell access to the environment running your LiteLLM proxy — the setup below involves editing its config file and restarting it.

Official Reference: https://docs.litellm.ai/docs/observability/mavvrik

Note: on first connect, only the last 7 days of usage are backfilled — earlier history is not fetched.

Configuration steps

Step 1: Set up the account

  1. Click + Account under genAI and choose LiteLLM.

  2. Mavvrik generates an Account ID for you automatically — you won't need to enter this yourself.

  3. Give it an Account Name so you can recognize it later.

  4. Click Next.

Step 2: Connect your proxy

Follow the below steps shown on the Connect tab:

Add Mavvrik to your config.yaml

YAML
litellm_settings:
  callbacks: ["mavvrik"]

Set the environment variables

Copy the three variables shown in Mavvrik — they're generated specifically for this account, so don't reuse them from a different connection:

Bash
export MAVVRIK_API_KEY="<your account's API key, shown in Mavvrik>"
export MAVVRIK_API_ENDPOINT="<your account's API endpoint, shown in Mavvrik>"
export MAVVRIK_CONNECTION_ID="litellm-<your account ID>"

Restart the proxy, then connect

Bash
litellm --config config.yaml

Once your proxy is back up and picking up the new config, click Connect.

It can take up to 24 hours for account data to show up in Mavvrik.

Troubleshooting

  • If the proxy does not connect, verify the Mavvrik values in config.yaml and the required environment variables.

  • If the connection remains unavailable, restart the proxy after saving the configuration and confirm that the updated settings are loaded.

FAQs

How much historical usage gets pulled in when I first connect?
Only the last 7 days — earlier usage history isn't backfilled on first connect.

What LiteLLM version and cost format does this integration require?
LiteLLM v1.91.0 or above. Cost data is exported in FOCUS 1.2, the FinOps Foundation's open cost-and-usage standard.

What does the Mavvrik LiteLLM integration do?

It imports cost and usage data recorded by LiteLLM Proxy into Mavvrik. Teams can analyze that data by dimensions such as provider, model, model group, team, API-key alias, and additional metadata.

Does Mavvrik replace LiteLLM?

No. LiteLLM remains the gateway for accessing, routing, authenticating, and managing requests across models. Mavvrik brings the resulting cost and usage data into the organization's wider AI cost management environment.

What data does the integration export?

The export includes cost, model, provider, model group, team, hashed API key, API-key alias, and additional metadata such as user ID in the tags field.

How often is data sent to Mavvrik?

LiteLLM exports the data automatically once per day.

Does an AI client get direct access to my infrastructure?

No. Mavvrik MCP answers supported questions using data already available in Mavvrik. It does not give the AI client direct access to cloud provider accounts, Kubernetes clusters, or SaaS platforms.