Keep your own routing.
Choose the upstream base URL, credentials, and path on each request. Constrain accepted prefixes. There is no separate provider registry or key vault to maintain.
Read the protocol guideself-hosted · openai-compatible · MIT
See every AI request.
Keep your own routing.
A self-hosted, OpenAI-compatible proxy with a live dashboard, SQLite history, and controls to pause, limit, or debug traffic. Request metadata, timing, tokens, cost, and outcomes; body capture is opt-in.
MIT licensedone Go binarySQLite or in-memory
request path
self-hosted
clients
apps
agents
IDE
millivolt
route · observe · control
openai-compatible API
upstreams
OpenAI
Anthropic
compatible
SQLite / in-memory
metadata and metrics
the short version
Route, observe, control.
One request path, one operator surface. The proxy stays out of your provider contracts and the dashboard stays close to the traffic.
See the request, not just the total.
Live rows, durable history, and scoped timelines for clients, providers, models, conversations, tools, status, and keys. Open a row for timing, tokens, cost, and retries.
Tour the dashboardAct before the retry storm.
Pause new work by scope, set provider limits, capture bounded debug bodies, export exactly the records you filtered, and protect against error storms.
Operator workflowsthe instrument
One view from total to request.
Global totals, scoped explorer cards, timelines, and the request log share one selected scope. The dashboard is dense because the traffic is.
01
KPI band
Requests, tokens, spend, cache hit rate, errors, and speed, with their ranges.
02
Scoped explorer
Drill into providers, models, clients, conversations, tools, time, status, and keys.
03
Timelines
Six chart presets share one time window and one selected scope.
04
Request log
Live rows plus retained history, with a drawer for timing and outcome metadata.
signal
The shape of the traffic.
See requests, tokens, errors, and cost as separate questions. A 429 is a throttle, not an error; the charts keep that distinction.
Swipe to compare charts.
operator surface
Control from the same surface.
These are operator actions, not per-user permissions. The shared token protects the dashboard and operator plane. Inference is transparent and does not use it.
- PauseHold new sends and retries by client, provider, or all traffic.
- LimitsSet provider-wide concurrency, request-rate, and token-rate policies.
- DebugCapture scoped bodies with a timed or manual stop; captured bodies stay sensitive.
- ExportDownload exactly the finalized records you filtered as gzip JSON.
- RestartDrain and hand off supported source deployments.
- SettingsEdit typed fields and rules with revision-checked changes.
run it
Three commands and a browser tab.
Linux with Docker Engine and the Compose plugin. No Go, Node, or source checkout is needed for the container path.
sh
curl -fsSL \
https://raw.githubusercontent.com/LLM-4-People/millivolt/main/compose.yaml \
-o compose.yaml
export MILLIVOLT_OPERATOR_TOKEN='replace-with-a-long-random-value'
docker compose up -dOpen http://127.0.0.1:8080/ and sign in with that token.
- Fetch the deployment file.The image is published for amd64 and arm64. Named volumes retain history and saved settings.
- Set the operator token.One shared credential protects the dashboard and operator plane. Inference is transparent and depends on the network boundary. With it unset, the dashboard denies every request.
- Connect your client.Use
http://127.0.0.1:8080/v1as the API base URL. Your upstream key is forwarded; it does not authenticate the dashboard.
Exposure is the boundary.
Keep the listener on loopback, or put an authenticated proxy in front. There are no user accounts. Read security before network exposure.
Connect a client.
Point an OpenAI-compatible client at http://127.0.0.1:8080/v1. Send your upstream base URL in X-Proxy-Base-URL, your upstream key in Authorization, and an optional label in X-Proxy-Client. The proxy does not register providers or keys.
Protocol and client integration · Adapters and compatibility
under the hood
One service, not a stack.
The proxy, dashboard, retained history, and operator API ship together. There is no separate database server, frontend runtime, or monitoring pipeline.
clients
apps
agents
IDE
one Go service
routingper-request upstream URL, credentials, and path
adaptersOpenAI-compatible, Anthropic Messages, Cursor Connect
dashboardembedded assets, live SSE, retained views
operator APIMCP tools and scoped controls
upstreams
OpenAI
Anthropic
compatible
stateSQLite or in-memory ringmetadata, metrics, settings
one binaryGo service with embedded dashboard assets
one credentialdashboard and operator plane; inference is transparent
two storage modesSQLite history or an in-memory ring
MITsource available, no hosted dependency
fit and limits
What it is not.
millivolt is built for a single operator or small team that controls the host. The limits are part of the contract, not a footnote.
One operator, no user isolation.
A single credential protects the dashboard and operator plane; inference is transparent and depends on the network boundary. There are no per-user accounts or tenant boundaries.
Best-effort accounting.
Sustained overload can drop records even when inference succeeds. The dashboard warns about process-local storage drops.
Observability, not billing.
Cost is provider-reported. This is not an independent invoice, a complete audit log, or a compliance system.
measured
Measured, not promised.
Current local measurements of the native process, including a deliberate overload fixture. These are scoped examples, not deployment guarantees.
33.5 MiBamd64 image
26.4 MiBidle RAM, empty history
114.3 MiBpeak RSS at 128 workers
14,166.2requests/s sustained for five minutes at 48 workers
Measured 2026-10-20; sustained capacity at 54d1276, artifact and five-second tables at facb699, on a shared WSL2 workstation. A separate overload fixture records the drop case. See method, tables and scope.
start here
Run it on your machine.
Three commands, one binary, your keys. Keep the traffic where you can see it.